{"cells":[{"metadata":{},"cell_type":"markdown","source":"## Strategy\n\n1. Split the train set in k folds [we have used [Chris Deotte](https://www.kaggle.com/cdeotte)'s TFrecords Id for spliting data]\n2. Fit a first stage model on k-1 folds and predict the kth fold\n3. Repeat 2) to predict each fold\n4. We now have the (out-of-folds) prediction of the k folds\n5. Split these out-of folds predictions in p folds\n6. Fit a second stage (stacker) model on p-1 folds and predict the pth fold\n7. Repeat 6) to predict each fold\n8. The CV error of the second stage is calculated on each predicted fold\n\n[reference](https://www.kaggle.com/general/18793)","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"![Model.png](attachment:Model.png)","attachments":{"Model.png":{"image/png":"iVBORw0KGgoAAAANSUhEUgAABUgAAAScCAYAAABZWmUvAAAeAHRFWHRteGZpbGUAJTNDbXhmaWxlJTIwaG9zdCUzRCUyMmFwcC5kaWFncmFtcy5uZXQlMjIlMjBtb2RpZmllZCUzRCUyMjIwMjAtMDgtMThUMTglM0EwNyUzQTI3LjY1MVolMjIlMjBhZ2VudCUzRCUyMjUuMCUyMChXaW5kb3dzKSUyMiUyMGV0YWclM0QlMjJvVjlnNllCamRfdWxKYUNIclc0QyUyMiUyMHZlcnNpb24lM0QlMjIxMy42LjMlMjIlMjB0eXBlJTNEJTIyZGV2aWNlJTIyJTNFJTNDZGlhZ3JhbSUyMGlkJTNEJTIyTFBNLXVsR2FzZUZ1cWtDNzQ1cU4lMjIlMjBuYW1lJTNEJTIyUGFnZS0xJTIyJTNFN1YxYmM2TTRGdjQxZVl3TEVHQjQ3S1RiczdQVE05MDczVHVYZmRtU1FiYlp4c1lEcEJQUHIxOEpBJTJCWWlqTUFJWkZ1cFZDWElXTmc2M3prNmR6MkE1JTJCM2JEeUhjYjM0T1hPUSUyRmFJcjc5Z0RlUDJpYUJzQWMlMkZ5RWpoJTJCT0lhUnZIZ1hYb3VjY2g5VFR3eGZzYnBZTktPdnJpdVNncTNSZ0hnUjk3JTJCJTJGS2dFJTJCeDJ5SWxMWXpBTWc5ZnliYXZBTHo5MUQ5ZW9OdkRGZ1g1OTlIZlBqVGZwcUtvb3B4ZiUyQmdiejFKbjIwWmFRdmJHRjJjem9RYmFBYnZCYUd3SWNIOEJ3R1FYejhiJTJGdjJqSHl5ZU5tNkhOJTJCM2FIZzElMkYyQWgyc1VzYiUyRmolMkJXZjBKYmwlMkYlMkZmUCUyRjJidWU4dVBQdDMxOSUyRmZFeG4lMkJRNzlsJTJGUUxQMmltaiUyQmQ3V2dWNFd2eXA0ME82Rk9aZkwwSDJ3bU9VRU9vZHZrRTE5bSUyQm5GJTJGRiUyRjYlMkZSdk1zc3lHJTJGaXdXbm1PaHolMkZxRHBGcEg4azdzNXZ3WjE1VzM0akhqcDhnRzlaS0gwWjczWGd4JTJCcktIRHJsJTJCeGRqRE4yM2lyWSUyQnZWUEloUGQ5JTJGRHZ3Z1RPNEdxeFV5SFFlUFIzRVlmRU9GVjl5NXZjUzBCRTloOExKemtadSUyQlB5ZVdtaiUyRjhPd3BqOU5hNCUyQkdwT1Vzd0xLTmlpT0R6Z1c5STNtT2J4SFNrWDZDa21YZ3VReW5DeUthREpUTWRnaXVKMVB2R0owUGlmbE5ZZDZLNU5TSGZ0ZnVtZUUza3l3b01KQ1ElMkZ1bCUyRkNhTWpYaDlVYkN1OTczeSUyQm5PSHoyN1R1akpoNU52ZDVPWUF1YlVtRElvbUtvc01kcTU3NGdhaHE5MndRNFJ1TUZva3k5UlliVkxxM1ZjNGt6akF2bjZJYmVtcnJXdVhtRjVETXJxWkdNaDhtSHNmUzlQVDF1eTlBbWZBeSUyQkJmRVpOclVJZHZiTHFVZkFTT2loOVYxRlBhNWxJc3lvVHhUQmNvN2cyRVY1bWVDamN0aWMzUkIwJTJCY0ZsJTJGeFA4Y1p6ekJJMSUyRlQlMkZvaXhMYjVpcUNZWUxwTkxIM1lSMmk1OWI3ZHVGelEwbVhSTyUyQk9EVndrWU00WWdXTVFTaiUyRmRHeVdYbHZoSEdxY3NrMWtPWHFOTGxrYVV1QWhRWnZ1YVFaWlNUWkZMbWtVRGd2SHh4Y01ObDJKOG5rJTJCRENLUEtlODdPak5pJTJGOG8lMkZQOG4lMkZsJTJCWkdlblZlJTJGTGRsZXppa0Y0TUpNUHc1MHlFUmJ2aGRwUUZyUXczbFZBRWFrWEdWQ25PS2hSQlZaOHkySVRpVUhJcmglMkJyWWlNTEFDQTklMkZGQzhLN3lLWHA3Y2xWJTJCTWpVYnN6SkZhM1oyMXNKTko4TnBNaFVSVUdoJTJCRE9jRGk1UktUNWtFYkY0VWtJaW9WRSUyRmM2UXFQY3pXSVpESXMyMzBSR0pWMjU5Z29vS3JsYWxBU3N4ell4NDJVU1pLajJ3OVZsN2pqR0M5VG1uZVQ4clFNRkEySk4lMkZZN2drUTBXTTBPeTBDSzlIbklhdGlKUnhzSEVJdlIwSzAlMkZjNGdlJTJGRGZlUWxreDN2MkhpJTJCJTJCeEVlZ3BjNGUweDJWYlhza0lwdHV6bk5zclBOT1lCbHkwNnBBVmNqSXdoakZNWkJXSGl6b3N3WGl3VUhRMUJYeTBTbGVqMXR0YzRGS2pjWDFaeEJObVFreDE4MDlxRCUyRks3YTI0VzdOUXYwNmRkMHcySCUyRk5CRDhaU0JnQmhSJTJCJTJCSThJUHg3RWltVk92V0J6czB4ZDl0TXJldXd6aU9OaW1GMkc2WFBta3lWSVpUJTJGZ1hMOTR6MlEwTiUyRkcyZThiVjZ1c2ElMkY1UFl3Zmc1MkdCM1FTeWlKWUJTJTJGb2lnZXdEVndodEhPQ2tNcURqTGRlbmdZMEZ5VkhUMUVPcFBqNmNjdGlTaHJ5aTl3aXk2SmQxd0N5aVFVRGslMkJnRERBTlYzNUMxNDNudW1nM0JBYjVBRWR2QjQ1aWp3a2NjeXpnJTJGSXhpJTJCUGdleGxDaVppalV2RlVRVTBDUnBZMEpJcHJYVzI1Q0UyMUNoeksxVyUyRmNraXhjcUdKelVkOHJQN1lZeGZSZmdSU3FMd2ZzclNYV0o2T1ZHT1FZdmxSUzlJNHRlUTJjVXZidzJaS3VESSUyQkMlMkIlMkJQbklMd0lwNEJabkElMkY3MlNDV0sxbXZSVENjcGVxY1Z2UmFqUjQ0Zkt1YVNuOCUyRnhpMGlpbDdQWmVudWtFa1gwMnB6dEZTbDZlNGhlVlpuYTQ1QTlUREkwbldFRThqalluTzNXMnlPVktCNEhXMFljQlpTOVlHcVhnODFRTEhHbkRDMWF6TSUyRm1iTGplSHFsRTBYdHpGWXVGZE01TDZCJTJCZVF1aDhJNkt6clFxanNwNmxSQnV6MXdxZno2VUJWbnN1RGEzWWkxOHFqYXJRVFBlT3NYQ3pIQXRYbkR3dDZUUUlBRkFVa3Q0NmZKV1BTUXZGZnlXYkZINjFHSHhmaHV3bGhwMmo5QmdWY1JsZDVRU3ZsT0VwTWdENjNucEhzSXZJRm84SENNWThCJTJGcnYwaGUyV0lBa1VvZ0c1MHF5Mk5CbFFGWEUwdW9UVlFwa3E1Vnl3eUUyNDM0bWNSQUdVWFNaSENBSmQlMkJzUXVxUnd0VlE0U243STNXbWU0RXd6T0t5JTJGa1VuZk04WG1OSWxSTFdBWWNQbWJxODFMbFhFRGMzV1RVQUZnc2FoeiUyRjZkUGklMkJHTCUyQjlvNEhIOGUwN0hRY3ZWUUwlMkJlRHlGbzVndk82b1N1dFdLT2xPM1BFV2dlelM0VGMzaFYwYmRlZ0FRTWlYUVZhbllSaTVmYUNMSm02TmJkWDRhYnRhWnh0S21scW4yRTFnYko3ODFKMm1kNHJ0djJZY3F4QXRyNnFEV0RVeUFUZnFYQWpqT05CNDV6TUtiZWlUbHVSS0RtJTJCS3BDWm8lMkJkWlJxQ1ltd3BrZkxRcnNVU0p1cW1BYzBxbkZNQjlCUERrbWI1NTRFJTJGeWRBUExpS1NNQXhrazdVb3NZVFJnWFRwakJCVEFrJTJCZjdxcnFNcHA5bkdaRUVzQzZ6czdzU1N4d0IzTUVGY1YyNUQ3UllBelgzUWVVV1hqSm9kbnhqS0hQWXlPVmlRV0tYNDZWRGtPMnFuQTF4U2U3RElQSFMyOGlJMEJXcnZodVBtaEdSZnduWnM1bklaMUNtajJucXN5eW0yTFZ4MXR4c24ydWczbG1XMHZnb3J1Mnp0QXdyVERGMnpHRG83NEJNJTJCTFJIb1ljZm5nVE84JTJGSFBwOEcyJTJGU2RwZkp3RzR0WHpYTXVZa09QYWFWJTJGa0dPTW5JR0xpVWJXcUJzTXhkVWRyNlBsdUxrM0RIRVpPR0RZb0V4U2NzRk1VRlRSVzRIZEVDRFY3UjFMN2N1MG15eTl1b1hiV2VtZ2thbmZKbjVIVVp0Y0JWSXVOdDhlbGRnZG5qVERVSG9LaTV4QXpBTFZObFluWVlGeEIzc0VESXd5eHI0QzFyVW9TWkNPMTU2TlN1em1oaGZXSWphcFZtdGxZcDhIVTJtbzBTZmNoNG1zVG0lMkZOM1R6WWxtN2ZSSm83MmtPVDRKODElMkZTNVB0ZHZWWmZFS285WEpic0Y2UDc4JTJGbVZGcW4lMkZkYjgyV3F6OWJHbUtTdDhseloyMWtVdDQwR2R6b08wODBjNFd0azBWNXprd1U0OE9CeTNDUHV0NjVQOGtPNTZUMEVReFlWamgwN2xTTCUyQmlkWWlpS0FndmtFMk8ybUc1cGFnVFJkVHBXUzZEVUxLT211RW1aWjJVZGN6ZnVvdUFmSThjTDhLcSUyRjljUTVmbjNFd3BIS1RNRmw1bG0xZjBpaHN5a09kS2x6T1F1UFQ0R2F3JTJGclZVNkJRMnZDZyUyRmdWcFBTUTBpTngzbXFBU1hxWTQwcVB5ODl5bHRKRGFseU1Ndk5MREoxdlZGUDBPVGxJYSUyQlVocVc1SmdYbDgxWm96Q1V4N1hIbkpFUCUyQjQ0SHc0STE4NmNUTmRUTHNjcWpDTSUyRnBrdXRia3NjOGFXNkRKWUlncGdTRk90WnpIZERqblZpbDVpV1AzSmFhdnRjJTJGR21wMDd6SVEzSG9TUVhyZSUyQnhvYUpEQVNoZ2xqWG1HZ0FOUU5GbVJubTZZbGh6TkVBTWNNanhuUUpDdDh6QjBLQmJSaWJhcHdRRHpWc2l3Y0MyVVpoRFNnZFZNNFNRRGd5WmlSSVFkT1hOdEdhS2FaOSUyQkJzT0dhYzVuUnVQTWs4Q0UlMkY3blF0d29UckJSV0JQJTJCRkt1Wk1LZnhvMDBPanVXSFVnT1U4ZlYwM1A2RndoJTJGemZQUFNLd21mOENQUVclMkYlMkZkejZHMWhlUGh4dHdwJTJCeDViOEhvV1BrZk9vSHF5JTJGM0xWQkZ2UEhmMzclMkJkdyUyQlBEJTJCMVpYNzNZUjA5JTJCNEh4TG5yRTJOOHAzZ2ozdjZUJTJGJTJGJTJCMWZVJTJGQkRJJTJGSVRqV1BJYzhvalFoWCUyRnRYUEx1elY5dnYlMkZ4U2Y4S0NoUFFVbUZEV0tiaWRGT2k2QmY4UEJnUHM2MFRxN1BiaDkyM3Jyclp6WDFDaWMycDB5aVhydkdTZkklMkZUaUJ2aUdqM0NKZkpRd3NVRmN1V24xWVZKaHI3aEpEYUlTa3h2akRlbTdsJTJGVnFUbDhoc1RvRlFXZUQlMkY2d0NmM0JaTUo0MHFicUw1V1owSVhaSG8xUkZvOHpMNDhQQVFTd2RwWmZRJTJCYlpPM1BLZlhtTGZJNjdLN3VueTllNnljMGhOZ0YlMkZxcHFIbyUyQlFlJTJGTUR4YVRxNlkyMVJ2dnpsdTNScExENFBiZGc2WDB3VG4lMkJnV2FlOFhSVEp1THU2N09jQXlkTk9PYTNUOUZtN3lTUUdwZjZBcHFubm9Tb0RBMHdKQkFhWTRpOEFGS0VsRVFDeWpVWmg0U0tHelJCWjBYVUhRTENBY1VHWWJpNDJpJTJCQkNmbkhjMlR3RVRHSSUyRm83bW5uQnhGYUZnOG5ROFFqMDVzVXBKa3c3dlU1QU1WTU5QYjAlMkI0WUpjSEFvWGhVcHV5Z0VqMTJEcEdJWTJVeFJ3bHNSNkhuVlElMkIlMkJFb2VVakZDcUxBa2p0NGFCMGJHbnc1NHh5YzlPVmx1ZldpS0hIQkxUeWZVbzdUN2toWk5ycFJidjBFSlRCdlAwRXBQOWRrbENPVU5HUG9QRWtwbjhBWiUyQldRWkE4a25VSlZQdGpLQmZPcmxkYnVUeG5MMkJjSHphbU01Mmx5Y0dzc1ZIc1czc1p6QjRPS1RIV291Ynl4bkcxUUglMkZjaU41WXdPaldvbHRmczNsbXVpOXJpTjVVd0daNXVrOXVXTjVScDVlMXhxZHpoQ1J4aHFYMTlqdVNaaWo5dFl6dFN1a05oWHdOclZ4bktOMUI2MXNaelpjcHF6TER1V2plVWVhSzZhNnl4bHJUU1dzMDA2RDQ3Yk9NU1VwZjhYOCUyQkJ3M0NMc3Q2NVBJaHZMU1ZIWEtPcXFqZVVFa1hYTkVROHA2NlNzazQzbHBNd1VxTEdjSURLenBhNU55a3paV0U1S2olMkJtbFI3V3hYSlAwR0xleG5DbVBEaGhNZWtpTnEwMW15c1p5VW1EMmJpelhJRERIYlN4bmRxdE11c0ZLcyUyRmE4bzc1dHlLYklZVEpsQVZGZktMUVVFRjJDakpZQ0lwdlN5cEEzVU9ZTU9RMFNLRlJxVnR1VFhZS01lbnV5U2NBZ2k0VDZieUJuNjFNdndVWkxmZW9rUU5Fa1VIb0M1WHcxMlNVNFNhcko1azI5cVNhQmlhd202d3VUZXR1eVFWWFNLY0RRcldic0JrMk1zdHhXbFF2T2FLOTJzNkJPeHAyaURFZnZTdmJ1b3k2b2lzcE5YNWdHS253YjE5OHlWRnJzMFl1ZzBtS1FUZ09WYnVWWEVpcEZLJTJGSnNTNHVMb05MUzAySWFxTWglMkJTbnpNa0l1UWN0NE9tUVlvMGgzYTN4QXBBMlUlMkJIRkNxZlMycVUwOEFGR3RvZDZnc0hEOVhPSjdTbUVmbHVLcFNrTWtaUGlxTG1wdVhFY0dsWDdGd2FjSElDSCUyRm1PSzBRSWhCd2dsME12VjBDQ1RXNTluMjRqN3hrc3VNZEc4OTNQOEpEOEJKbmo4bXVxa1ZBam9PTTFhb1dXQ1d0TWt4Z0E3Y2VESzBWSVVVSVl3M0dRVmg0czZMTUY0c0ZoOWhwdFlKY3kzTFNpajA2S1dCV3VkVVNxU3pxYXQ1SUZaUFNnJTJGNnZ5SW5oYnMxQyUyRlRwMTNURFlmMDFnbkE0a3BmQW8lMkZQQWRrWXI0NHhnbEpCNEglMkIlMkZSRkg2Mnk5eTZET0E2MjZVV1lybGMlMkJhYkpXeGhQJTJCeGF2M1RQWTEwc3I0R1YlMkJycDJ2OFMyNFA0JTJCZGdoOUVCdllTU0NFYnhLJTJCbCUyRlhBUlFUVGd5SWVJY3A1MlZhalFnWkcxSU9PQ0FwbUowVEglMkZSbWZvY3A5VHZrb1F4SUI1M08zTDNDWThCSnQlMkZLVDBpNjhWd1g3WWFBSHlmTXpBWERqTldoeEZqS0R1Nnk0MUNtZHhzc0xHNndZQWliM0NrTFclMkIzSzdNaTBZdkNCU3hZZW00V3olMkZvVFRTWGJPV3VFMXM3QWgyaTdjb1ltUFpPR3hXTmhpcyUyQno0d2NMbXJKeGRNd3ZiZ3JHd3pWbGpraXpjaDRWVlpXcE4ydWFzblYweER4OVpSaUJOMnBhT05CRjVHRXl0U3R1YzFiTnI1bUhCSEZvZ2klMkZWZVMlMkZ3RHFhNkI1ZyUyQlUlMkJJZHR6Z0UwSHdTTGY1aWduQk5LaTM5WW84WSUyRmdDSlZyd25FZHNwcFoxbWZCZ1NPck4lMkZjejJ2ZyUyQk1mbkVMbWVFNWViRU1nWUNDTnUydFclMkJrWEVqSGFnaXlZJTJGbUdBZ1ZGcnlzQWFCMHlLUzROeFp1ZDZDT1RDdHB1UW5Jd3BRWXlNaVN2Y05SQ1BmR3d1MkcyOGkwa2pFUUFWbVlFZ01aRnhacWh4TVA3bzJGMjJNZ0k5TktKaE1KeU1LMEdNaTQycG5hd1NkM1p6eXN0bWNUalV3cjZVd1RrWWNwTVpDUlpUdEQ3Znk5OHJCb0RpMjFnME5MaEJqSUNycTJhenhRWWlBUTZTclFIZ1NQZ1lDc3kwaGJERVRoUjNMT3ppb3B0czl3bWtpczM5eDFWc1pBQk5veTJtdUhSc1lOWjAlMkJabEIlMkJkNUljb01SQlZacjZjNXhpQkxEZU5zNmRNc25BZkZwNDhCcUp4ZHNwZE1Rc2ZPVWFnWFRnampXUmhrVmg0OGhpSUpxc3h6M09NT0N5c2RTbTdjOEtBbklEUWRrSnZaZVhZRHVFbFAlMkJUdTFHVkQlMkJwQU03d2d4OUdySEU0MTZrb0JHSXdHM2t3UXNtaTFUT0htbDVheU9ZWThLSVQ4THlsRWhuejR0MmclMkJzb0puVGx4eFhRWTR1Y1N5MFhEM1VmSEF1Uk5hS3RCRVMlMkJlQUtZMzZtdzA1THlqbmdoN2ZtdmhzVGdldkx5M0xyUlJGeHpTZ0x6NmVjd2RjSmUzZUdNbUN6b1V3RjVwZ3dzenZZZDg3QjklMkZBS2hWcjc3ckk4cnVYSFpUNEFuVyUyRnJaSVUlMkZ2Y1I0R3BRUk9UMGozamhQZ1ByJTJCMUwzajB3QkVmTXczbUV5TG94Q1F0aSUyRmwlMkZSUTRVTENES1NZeUJWY3JCJTJGJTJCTVRrR3NXODBZaVdqem95SE5ST055eHBpcEtPUUVTdVl6eHZvJTJCUG0zcFJLd3QlMkZPcDdHTU1MdHdzMnolMkY5dEhxMzFDRUFadEtwUnR4NVZsUUpaZmlkclVVdlRHJTJGWGhZVkc4V0JBZFpUd1VFemRCR2NTWFFQYU96NGlyQVRuUEtwZ1F5RjFzNjVmUVB6eUZlQ2NrenExTFRHeXp1cnBoRU1NazdnbmVQNnFXTXRCNlclMkJYbHRobXQ2RG0lMkYxUjZ6dGZWRG9WOXhvWDF4UThmaVUwdGI5ZUhVMERaOVg2ZDJ0c2R2MU42OEZuJTJGcnBCRXNDenpqektQYWRPZWs3WEFmMVlwV1ZlVlg1bmJzbFkzT3JzekR1MUV5UzFXJTJCeEdjZmZMYjdWTWZFNTF5ZkZackUyNVg0S1hOTCUyQlBJSjglMkZsUjhpT2hGU2dNMXJ0RWF4JTJCMHRtZGY4a1NyVmdhcllROEIxcXlKZHdaV2M5d200a0JoY0ZSSXNQWUJxeTRTV0sxQkpHc0ZyRFdObERkWXMwUE43aUxpWmxaMlJ6MUwzSnNxMmdhMERzbm1ON2Y4JTJCUUYwa3dVN3V6UyUyRkVTSFglMkY5clBld0FVaWxQUGUlMkJDVzY2OVIlMkI5dzBrRnhtQ1EyVmFxSXg5SzBadGRXWlJ1MWJJODk3RUNnOUtXVlZrVEFqNjRSRWtoMkNuUGVnOGU2ZmM4MHMzUE9ZSDM2MGtxVTZBckx3MU9jOWFMejc1MXd6QzdlWDZveExLOTVOYlNRTDkySGhxYzk3MEhqM3o3bGlGbGJiUzNYRzdUT3RpWmVMWUElMkJVaWpDM2xEemROWE16bXZhTTBkVmxhRE5leVhkQXU0Yk0lMkZqOWtYbjlYeE9tVnZINURyWXZpJTJCYmhlN1E3YTFKV3h0MkZVUEpvR293JTJGYjBuaXRObURRaHdyeHhuUmZZZzgySGljVCUyQklqcWlwUFpNQ3BMelJ5THMxc200aDJNQTVxa1pKa0FQYk9yRE1PZUdlWGo2QTFsNURRQXdCQm91R1ZpMXJtcEp6SG5sajJ6dFB5SSUyQmV5TWtweXVOWldKTzJXYlF3ZmlLRkolMkZTa1dxNjk1dUtXVzluYVpJWmJRZlI1RUNIYnlWSXRjOGlWNjFabkF6dWtBWFZWaFNzRGNGVlc0VTFEdTRHeVVGZTFOd3pvJTJCQU5NZGd4NjNZWUlyMEx3TGZKZlVKc3E2dkFvcThMUGhRUVVDTEtzcXRIaHpvTkZPTEh5ZzBDWXAyVUpnYXBVSjVaRmdNY0FwYUIxamcyMyUyQlN5S2dqdzg2UlVLMnpLb1lzS0tvNFIyUjBLJTJCaExsOW1GMFNaM1Z4YklSTVklMkZ3eGd2JTJGUzRaSWRFMW1wRlBXVnB4TGY1SDFWUm1sYVpHTllxd1d2MlBhc1VabE1kOXh6THpEWnJpTjZsTmoyV0ZRbVFHa1IzNHozT1NFSzRwdjJGbWRtR2xtYkJzVEZPUktsbDhPSzlocXdtVUhLdmpHUGRHczE0cThYVjklMkJIcFVLNkVoMDJobzU2YU5pckptUlZlaTdBWlFacyUyQjVvZ3hmaGdFaDMybHJ4Vjl6ODNQZ0luTEglMkZ3RSUzRCUzQyUyRmRpYWdyYW0lM0UlM0MlMkZteGZpbGUlM0U67RreAAAgAElEQVR4nOzdeXwU9eH/8XetVn/229ba2uP70Gpr7RdvqiQkmkAsWqFVCy2ooFYOCWc4KiIoykK4scGDenAZUG5UBBIuuUGDIiinBgGrgFgrKIgo1+f3x2bC7Ozslexmdjev5+PxfqjJ7MzszGSFdz4zHwkAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAC1U9eiZYYQknzx+rMBAAAAAACgVuhatMyU7ztOCEmiUJACAAAAAADUEApSQpIvFKQAAAAAAAA1hIKUkORLRUGaVxEAAAAAAAAkSiIK0hZ3GnPttcY8OvhkwNdXrj9urr3WVKa627Gvyy2xrGtmyYmoXvfo4JPm2muNaXRj5PU73++11xrT4s7qv++aSKMb/fvbufvJsN93nuNoj22oayTUtRRqP0Jl9LgTAeco2vObLKEgBQAAAAAAqCE1WZDGsxx1ri/ZClJruVBZud77Ei7a/Q93rKJ5HzVdkLqdIwpSAAAAAAAAuKqpgtQacRjPctBa38ySE9VeVzwLUvu67MfA/vVkH0lqH/3qPL6xjKKN5diGu5aqW5CmWihIAQAAAAAAakhNFKSJKEfL98VWkI4ed6qkcysoQ5V4nbufGknZ6EYTVflmvcat1Ht08MmAbTvLVOdxcrtN37lO+zpCHedolnEm1G32brfXW+fcfqzCHdtQI0jt10rn7iddC1K39zJ63InK4+v83qODT4Ytae3bdLue7F+3L5vIkpuCFAAAAAAAoIYkuiC1F2fxGOlZlYLUXnKGKhvdCrRwrwtXkFolWjTv163ss9btLHXdSjy3AtX5PqJZxi2RblW3CtZQx8kqNaMtSJ1Fpdu6wr2XletjK0jDrcu+X6GWcSt4KUgBAAAAAABSTCIL0lhKxeoUpG6xRhTaizHra/bXhivxwpVl4d5LLIVwqNvx3fbRfmydJap9f6yi0Xq/0SwTad+s92IVkPbRk87j5ixWoylI3bZl/5p1DKx1u50T5yjSaEex2pdzG8Xrds6t1yZqFCkFKQAAAAAAQA2piYLUXjqFK+QSUZC6FXpupaGzQAt1S3Y0t9hXtSC13/Ieavv2UY8r1wePQHUr7KJZJlSct9OHm73eOSIzloI01HkK9wxS58jVWAvSUOfJWR67Xbuh9peCFAAAAAAAIMUkuiC1yid7mRVr8RmqVI2miAz1PNBIhWiogtJtNGao/YqmDK7KdkKN6nTGflyiWcYt9rLRWc66ne94F6TO8+f2SIKqFKSh3ovbvoU7nomaCIqCFAAAAAAAoIbU1Cz29kIq0ozk8SxIvRhBGm7U4+hxJyKObAz39XDFnrMIDTW6MZpl3PbDKivtr7E/J9VZ2CZiBKk1utPtFv9EjyClIAUAAAAAAEhDNVWQlu8LHEUajwmbolmXF88gDfVcUfvXrYIv3OzqboWy87mZbqWls7yLZplwcU6e5HarudvzORPxDFLnPridX7fRt/F4BikFKQAAAAAAQBqqyYI02oIxngWpvSB0S7gCLdRt6dHsf7jX2su3cAVpdWext45/tLO1R/teQj0rNdQxinYW+1C36tsL0nDLOAtS+3uMxyz2FKQAAAAAAABpqKYL0nhO2BRtQepWmjlv5Q5VVNrL1UY3mpiKMbcCLtrthltHuGMRrvisSjnq3Ee3W/Lt59Q6LvYyNdqCtHxf4GjVzt1Puj6uwFmKWtt3G2lrfT3ccXaOkHVeTxSkAAAAAAAAaSwRBSkhpHqhIAUAAAAAAKghFKSEJF8oSAEAAAAAAGoIBSkhyRcKUgAAAAAAgBpCQUpI8oWCFAAAAAAAoIZ0LVpmCCHJF1GQAgAAAAAA1Kg8QkhSBgAAAAAAADUgjxCSlAEAAAAAAEANyCOEJGUAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACctszQ97aNUa/ySeeWvD/hewc+mHzBjm1jZUhy5YPJF+x4f8L3DpRPOrdk21jdb57TGV5fOwAAAAAAAEBKK5/0497vjf/ut/sW33HoyNZJ5tjOBcbsXUeSNMd2LjBfb51k9i2+/dB74047vn3iD/7h9TUEAAAAAAAApKTtk86b8d+VXfZ7XfqRqufzVV0+3/7CTyZ6fS0BAAAAAAAAKeWDF3/50mcrOn7tdcFHqp//LGt7qPyFn0/1+poCAAAAAAAAUsL2F8576LMVnT/zutgj8SxJO/y3vPjcXl5fWwAAAAAAAEBS2zJD33tv/Bnfel3okfhn27jTjhmfTvP6GgMAAAAAAACS1tZxpz+477WW33hd5pH455OFt38726dhXl9jAAAAAAAAQNL64MWfL/566wuel3kk/jm8pdis/ddPXpd0gdfXGQAAAAAAAJCU3p9w5hfHdi7wvMwj8c/RnfPNlnFnHpTU0OvrDAAAAAAAAEhKOyZfsNPrIo8kLlueP/8TSQ0kne/1tQYAAAAAAAAknW1j5XmJRxKXbWNlJOVJusrjSw0AAAAAAABIPhSk6R1bQZrj8aUGAAAAAAAAJB8K0vSOrSDN8vhSAwAAAAAAAJIPBWl6x1aQ5nl7pQEAAAAAAABJiII0vUNBCgAAAAAAAIRBQZreoSAFAAAAAAAAwqAgTe9QkAIAAAAAAABhUJCmdyhIAQAAAAAAgDAoSNM7FKQAAAAAAABAGBSk6R0KUgAAAAAAACAMCtL0DgUpAAAAAAAAEAYFaXqHghQAAAAAAAAIg4I0vUNBCgAAAAAAAIRBQZreoSAFAAAAAAAAwqAgTe9QkAIAAAAAAABhRFuQSgqb3W+Xui47c8zwoK89O/yhyn+PZtu73y6Nafl0jXUcmjbOoyAFAAAAAAAA4iHeBWmP9q0Cvl42rzjoa3MmjkqqgrRp47zKIjeZ1920cR4FKQAAAAAAABBPsRak9pGi4Uo8eyno9rVkSSL3LZ7rvqLOxZXngIIUAAAAAAAAiJN4FqRWIWjlijoXu34t1IhQ57LWtkIt7xyZ6iwira+XzSt2Xa/bvkXzumi2H2rd1SlHrX9SkAIAAAAAAABxEo9b7Hu0b1XtgtS5nP37bss7y0krRb6eEffZKisjFaShXhfN9uNdkJbNKzYzxwyvakE6VVLTuF44AAAAAAAAQBo45+VH4leQ2ovBcLfYOwtP+0hNZwFZ5OsZtLz1387S0bkOOUpLt+247W+k10W7/Xjfvl+VgrTLrZXn6StJ90s6PaFXFAAAAAAAAJACLpL0vKQDE3vV/DNInYWnVfzZi1Z7Qi0fKmXzikPus/Nr4QrSUK+LdvuRClLnLfyRRpnGWpDu37rUnPW9oP3bJumPCbmqAAAAAAAAgCRXV/5itLIwW/tE+hek4d5HuII01OtSpSA1e9eZVx6VkbTVZT9nSvpN/C8xAAAAAAAAIPnkSVqm4JLsQy9msY/lFvse7VtFfYt9qH2Od0Ea7faT4RZ72yRNz0ja7zj/30gaIunshF15AAAAAAAAgIdaS9qg0KMdi+MxSZMijJqMVJDal3HGXkhGs7y9tHS+JlxBan9trK8LtX23dXtUkDaS1E1SE0kvSjrq2O+PJN0h6TvxuOgAAAAAAAAAL50j6V5JuxSh1JTUI1kKUrfS0SojQy3vnEneWRy6vca5bvvo1VgK0mi277ZuD0eQNpF0TcW/t5S01uU8lkm6Oi5XIQAAAAAAAFDDzpHUX9IBRS5GreRFW5CS1IytIM2ruE5+Iem6iv/uLenfjmvihKTnJJ2b8CsWAAAAAAAAiIOLVDEjvaIvRq2IgjS941KQStJ3Jf1aUgP5b79/WtJXjmvjc0mdJZ2esCsXAAAAAAAAiJNz5C/AfJKWS/pCkcvR5RIFabonREFqOUvSVRXfayppoaSTjutkm6SGcb5eAQAAAAAAgIS7SNIShS5IH5coSNM9EQpSy7mSMiuW6SRpq8v1MlPS+XG+RgEAAAAAAICE6ajwI0hbSxSk6Z4oC1LJP4P9BZJyKpYdJmm/45o5LOkRSWfH7zIFAAAAAAAA4u/3ko7qVLH1jKRDCiy7LpIoSNM9MRSkljMk1ZH/tvom8o8cPe64dj6SdEccrlMAAAAAAAAg7s6Tv8CyyqwVki6VNMv2tS+shSlI0ztVKEgtP5R0TcXr/i5prYJHIS+XdHXVLlMAAAAAAAAg/s6QtEqnCqx/S7pS/pLrTgUWW5IoSNM91ShILb+QdF3F63tL2qPAkvSEpOfkf44pAAAAAAC1z6UNGlxyTVbWXzPqX/dcRnbOS9dkXV+akZn7WmZ2zsrMrJw36mXlvF0vK3djZlbO7ozsnL31s3M+zczK/W/9rJz9mdk5B+pn5ezPzMr9b/3snE8zsnP2Zmbl7K6XlbuxXlbO25lZOW9kZueszMjMfe2arOtLM67LnZVR/7rn6tW77i+XNmhwidfvHUhCzyrweZHXS2qgUwXZSxXf81kvoCBN78ShIJWk70r6rfy33d8k6WlJXymwKP1cUmdJp1VjOwAAAAAAJIerc3Jy6tXPmegvNq/fn5md83VmVs7RzKzc4xlZOSczs3JMRnaOyczO9TwZ2TkmIyvH+Pcr93hmVs63/v29fn9mVu67GVm5xXXrNrjO62MK1ADnpEz3SMrSqXKsrvzPHbUKM0kUpOmeOBWklrMlXVWxrqaSFko66bju3pG/SAUAAAAAILnVrVv3onrZOZMy6ue8k5GdeyAzK+fbipKxSkVlsztbm1Ztu4aMr2h81Am3nmZ3tq5amZqVazKyck5mZOV8m5Gde6Beds6GjOzri+vWrXuR1+cCiAPnpEyPV3wtryJZ8t9+L/lvrz/HeiEFaXonzgWp5VxJ9SvW2UnSBwp+PulMSefHcZsAAAAAAFRd3bp559TLvG5KZlbOp5nZOceiKUEb39bctGrb1fQufKKyuBz9YqmZsmi9mbJovZm9utys22M8z+zV5ZX7NPrF0sp97V34hGnVtqtpfFvzqMpT/3HJ2ZeRef2LdevmnRP5qAJJw21Spst0qhTLlfQ/tuXr2l9MQZreSVBBKvlvpb9AUo6kGyQNk7RfgSXpYUmPSDozztsGAAAAACC8+tkNhmbUzynPzMo9kpGdczJUMWgf3TlizHQzZdF6s3DDXs9Lz0QWqSPGTA8YpRry1v2snJOZ2blHMurnvH9t9vVDvD6nQAjhJmWy8rNwK6AgTe8ksCC1X4N15L+tvon8I0ePK7Ao/UDSbQnaPgAAAAAA0jXX5NySkXX9pszsnKPhytDehU+Y0S+WJs3oz2TJ7NXlZvSLpZWjTkOPNM05mpGVs/Gaa3Ju8fqcAxUiTcr020groCBN79RAQWr5oaR6Fdv5u6S1Cr7tfrn8ZSoAAAAAANWXmZndMiM7Z09mVvAI0WZ3tjYFfQabEWOmU4ZWozQdMWa6Kegz2P25p1k5JzOyc/ZkZma39PpaQK0VzaRM34m0EgrS9E4NFqSWX0i6rmJ7vSXtUeB1ekzSKPmfY4owCud2uKRwTvtbBs5p32toacG0fy7qVTJsfsEHQ0u7fjK4pPMXg+Z1/KZoce8dg+Z1/GZwSecvhpZ2/WRYacEH/1zYq2RoacG0AfPy7y+c0/4W36vtIv6iBAAAAABSxrVZWfUzsnL+nZmVG1CKNr6tuSnoM9iMfrHUrCw/5Hm5mI5ZWX7IjH6x1BT0GRz8XNOsnJP1s3I+vDYrq77X1whqjVgmZQprx+QLdnld4pHEZUvx+XtVswWpJH1X/tHLDSXdJGm8pCMKLEo/l9RZ/meZQpJvbn7OoNJOA4bN7/ZO0aIHdo1Y0PPgMysKv5j61jNH522ablZ+sNis3rHUrP1wtVn30Vrz7p71Zsu+zebdPevNuo/WmrUfrjardyw1K7cvNvM2TTdT3nzm6DMrCr8YsaDnwX8ufmDn8AXdNwwu6ewb+Gq7671+rwAAAAAQs4z614/MyM79xlmK9i58ghGiHmX26nLTu/AJc/OtgWVpRnbuN5n1rx9RQ5dGniSfI/3kv721Xg3tA2perJMyhfX+hDO/OLZzgedFHol/ju6cb7aMO/Ogar4gtZwt6aqKbbeQtFDSSQUWpe9IqrW/XBo0t+Ndjy28/5WBc/NPPLHk4QMz1o05trx8gdmyb3Pcs7x8gZnx1phjTyx56EDh3Pzj/1zYa9agks7cBQEAAAAgudWrX/+mzKzcI1b51uAPjU1Bn8GUokmW2avLTUGfwSb3hpvtRenX9erXvynBl4hPwc/4s2evpPsSvA+oWdWelMlpxwu/XPT11omel3kk/jm8ZYJ5Y/RPyuRdQWo5V/4SNE9SJ/knbbJ/Vp2UNEnS+R7tX40aVNKuweCSLpMHzs0/MXbV0P3L3l9gNn+yMSGlaKhs+mSjWVY+34xZOWR/4dz844NLOr/gm9s2x+tjAwAAAAABMrNy3rXKttw/3Gx8ReO5fT7Js7L8kPEVjTe5f7AVpfVzNiTwMvEpfEFqZXgC9wE1q9qTMjm9N+H03vsWt/zG6zKPxD+fLLr96Au99JROPXbBS6dJulD+Ec43yP8c0i8V+Fl1WNIjks70aB8TalBp5zuGlHbZ9eSSh/eVbJpxrKZL0ZBl6d53zbyN0489ueThT4aWdt0xaF6nFl4fKwAAAAC1XN26eedkZJ2ajZ5iNPViFaX22+7r1s07JwGXi0/RFaRG0p8TsH3ULOekTO1UhUmZnMxzOuO9cacd97rMI/HPtrE6efrpuqHi+kiW0YFnyj+TfUNJt0qaKem4Aq/tHZIae7WD8TZoXqd7h5R23Ttm1eDPX9+53PNCNFzW7FxmnlsxaP/Q0q57BpV0vMfrYwcAAACglsrMyj2emZ1rmt1xL7fSp3hmry43TW+/p6IozTmegMvFp+gL0h1iMpRU5pyU6VlVcVImN+8//4N/fLqsw36vCz0Sv3yy5L4vJvc92xo9mif/s0CTyQ/lf1ZynvzPTX5HwZ9b8+UvU1PSY6U9mhQtfnDj+NUjD639cLXn5WcsKdu1yoxbPfxQ0eIHNw5dUHCj18cSAAAAQC2SmZWzPzM71zS7szWjRtMkK8sPmb9UlKQZ9a//b5wvGZ+iL0iNpII4bx81I66TMoXy9nPnztq3tO0Rr4s9Uv3sW3LvkSWP/bjUdo00VPI+3/N/5X9URJ6khyTtUeDn1jH5b8f/gUf7FzPf3Pyzh8/vseCppf2+ePPfr3tedlYnb/57jXlqab8vRizsWeJb1vosr48tAAAAgDRXNzu7hXVLNiNH0yuzV5f7C9KsHHPZZVlN4njZ5Mlfkj4uqbgiL0n6Ru4F6WJJP4nj9pF40UzK9Is4bes7K5/46WufLu902OuCj1SjHF2af3D5qPMWOq6RvDhdI4lyuvzPz20o6SZJ4yUdUeDn1z5JbZTkI+GHzu/eeVBJx6MLtr50wutyM56Zv2XWicJ5HY8NX9Cto9fHGAAAAEAaq5eVuz4zO9fc162v54UeiX9ate1qMrNzze/qXL43AZfPRQosQobJvSDdLykzAdtH4kSalOl3cd7e+dP7nf3ktrHfPb5n/t++/nprsTm6c77npR8JnaM755vDWyaYPfP/enjbWJ103FafV3HNJOvoUaez5X+Wbp6kFpKWK/hz7B1J9b3ZvfBGLOi+bOa6cQe8LjMTmRnrxh4YMb/HEq+PNQAAAIA0ZU3MVNB3sOdlHol/CvoONpnZuebKuteekNQ+zpfPRQoeLeYcfWWlhaSfxXn7SIyETMoUhTpnna68SQ9o9OtP/fj1LWPPPLj5+fP3bhsrQ5IrW4rP37tl7JkH1zz149df6KWnbBMyWWkg6f8ScI0k2k/kL0HzJHWX9IECfxZOSpqkJPks871631WF8zoeXrzt1aNeF5g1kUVbXzk6aF6nQwPntL/c62MPAAAAIM1kZvsnZ7r1b608L/NI/NP41uYmMzvXXH1Nxkn5Z20+N46Xzznyl6T2uE14Yj2H9NI4bhuJkaMETsoUhcvkf7ZpHknZ5Ci1f9ZPk3Sh/NfhDfI/h/RLBX6efSnpfklnRrlOn/y/WIiboSVd7nts4f2fb9j9tufFZU1m/e63zMiF9+8fsqBTm3geTwAAAAC1XEbWdV9nZPEM0nTMqWeQ5prLr/j9UUmvSIrns0jdzJJ7Qdpf0nUJ3jaq5zxJ/1GCJ2WKwvnyPxOSojS1Yp2vVLmtPpIzJV0u/3u6VdIcSScU+Lm2Q1LjKNblk7QhXjs2pKTr0Klrn97tdVnpZV5c+6/dQ0oKCuN1TAEAAADUchn1cw5kZuWYzOxc06xlG2axT5OsLD9kmrVsYzKzc01mVo657IqrrYK0Q4IvqQlyL0iHy180RDviCjWrJidlitb5FfuQo8Bb/EnyJEv+83Ol0qcYdTpHUj353287uY+Sny/pN2HW4atYzlfdnRk6v2DS1HXPfup1QZkMmbx29H+Gzu9WXN1jCgAAAAD+gjQ7t/JW7GYt2zCSNMUze3V5ZTlqnVdbQdo5wZfU43IvSEfLXzD8KMHbR9U4J2VqoMROygSkmv+Vf+KpPEkDJH2qwM+4byQNkfQDl9dan4sH5H8USZUMn9/txVfffXG/18VkMuXlDRMPjFjQ4/mqHlMAAAAAkHSqIH142DPmvm59/SMOs3PNwFETGE2aYllZfsgMHDWh8hze162veXjYMzVdkBbKvSAdL3+xcF6Ct4/YuU3KdJ1OlaOJmpQJSDWny//LgoaSbpb/c805Md0+SW0U+DOz3Pb9ZVXZ8GML7585a/2EQ14XksmYGevGffXPxQ9Mq+pJBQAAAIDKgtRXNN6s22PMiDHTKwu2ho2aUJSmQKxitGGjJpXnbsSY6WbdHmN8ReNruiD1yb0gLZY3t2kjPK8nZQJS0f/I/4uDPEktFFiAWimTVL9ieef3e8SysaGlBRNffffFA14Xkcmcl98uPjhsQcH46p9aAAAAALWSsyBdt8eYhRv2moI+gwOK0m59h3DrfZJlyqL1plvfIQHFaEGfwWbhhr2VyyRhQZquzylMRckyKROQqn6mU8/H7S7/s3vtn3snJU2StMbx9ahvtR9S2nU4zxyNLi++8dTnQxd0GxTncwwAAACgNnArSO0FnL0ozczONU1ua2EeHPQkZalHmb263Dw46EnT5LYWAeeloM9gM2XR+qDlk7AgvTDB20d0knFSJiAVnSb/51qupEaSRkn6UoGffycU/JkY8Vb7ISUFHabU8tnqY80Lrz+5b8i8ru0Sc6oBAAAApK1wBamVhRv2Gl/ReNP4tuZBZWm3vkPM6Mml3IafoKwsP2RGTy413foOCSpFG9/W3PiKxgeMGE2BgvSiBG8f0bFPynRUTMoEVNeZki6X/+fnVklz5F6M2tM61Mp8r9531WMLezEhUxUycn7PL32lHS5L6NkGAAAAkF6iKUidIxh7Fz4RVJZmZueaZi3bmG59h5iRY2YwwrSKmb263IwcM8N06zukciZ6Zynau/CJqI8vBSlcOCdl6iwmZQLi5RxJmfL/LL2q8AXpgYrlgwya1+nrDbvf9rxsTMWs//gtM2hup8M1cK4BAAAApItYC1J7Fm7Ya0aMmW7u69Y3qMizcle7AvPgoCfN6MmllKYuZejoyaXmwUFPmrvaFYQ8hvd162tGjJkedqQoBSmixKRMQOLVlbRJ4ctRK684Xzx8Qbeli7a9eszrojGVs2Dry8cHv9p1ZY2dcQAAAACprToFqTNTFq03I8ZMNwV9Bptmd7YOW5re1a7ADBw1wYwcM8NMXbwhbcvT2avLzdTFG8zIMTPMwFETKt97qGPT7M7WpqDPYDNizHTXZ4pSkKIaIk3K1EBMygRUxzmS+iu6YtSeptYKRi78R7cZ68YyY30cMqXs2YMDX+7YrWYvAQAAAAApKZ4FaajS1Fc03tzXra/rbfnONLmtReWo04GjJpiBoyaY0ZNLzdTFG5KqSLX2Z+riDWb05NLKfbVGgzqfF+qWxrc1N/d162t8RePjUoZSkCKMaCZl+l9vdg1IC3mSdin2crTyVnvf3PyzB83reNTrYjGdUji3w3HfjBbfq/GrAQAAAEBqSXRBGqo0Hf/KSuMrGm96Fz5hWrXtGlV56jrismWbylGZbrGKy2gSbj1uzwONJo1va25ate1qehc+YXxF4834V1YmrAylIEUYTMoEJFaepB7yfwZaWe5I2Fvth83vPm/+lpdOeF0qplNKN8089uiM/OU1dREAAAAASFFeFKThsnDD3spb9X1F442vaLwp6DPYtGrb1bRq2zXsrfs1mWZ3tq7cp4I+gyv31bo1virPC6UgRYIwKROQfM6RbQT37zJ+NWzY7Ps/97pQTMeMWtT3y7ZDb2nm6dkGAAAAkNySrSCNNbNXl5spi9aHjFVcRpNw60mWW/spSBEjJmUCUsBji+5/861/v+55mZiOWfvhGjPwpS5bxS+CAAAAAISS6gUpoSBFSBeISZmApDdgTvt7xq0azujRBObZ5YO+bDPklt5en2sAAAAASYqCNL1DQVprnSFpvZiUCUh6Q0q77Fn74WrPS8R0zhu7VpqBr3b6VNKPvD7fAAAAAJIQBWl6h4K01mJSJiAFDJiX32LMyiH7vS4Qa0OeWTbw4F2P3ny/1+ccsPFJ6h8iAAAAqEkUpOkdCtJayTkpU1elyKRMW2boe9vGqFf5pHNL3p/wvQMfTL5gx7axMiS58sHkC3a8P+F7B8onnVuybazuN8/xHNuqGlLa9cPXdy73vDysDVmzY6kZOLvzHkk/q+5561q0zJDamTj82Nu5/XnJSrV5fayS+LgDAAAEoyBN71CQ1jpukzJdqxSYlKl80o97vzf+u79mhqAAACAASURBVN/uW3zHoSNbJ5ljOxcYs3cdSdIc27nAfL11ktm3+PZD74077fj2iT/4h9fXUKrxlbRr8NTSR3j2aA3m8cUPHfxzh+vyq3vuuhYtM+X7jpNalkQVpP3796+M4lyQen3MkvS4AwAABKMgTe9QkNYqKTsp0/ZJ583478ou+70u/UjV8/mqLp9vf+EnE72+llLJkJKuU0o2zzjpdWlYmzJv07STD05svUDST6tz7tKleCKeF3VGkrHTqf+HX6Rq/pkpXa5TClIAAFAjKEjTOxSktUbKTsr0wYu/fOmzFR2/9rrgI9XPf5a1PVT+ws+nen1NpYqBc/NPbP5ko+elYW3Kxr3vmMI5+cflH01fZelSPJEqFXV5FYmHcAVptbeTLtdpAo47AABAMArS9A4Faa2RkpMybX/hvIc+W9H5M6+LPRLPkrTDf8uLz+3l9bWV7AbMua/l2FVDmZzJgzyzrPDQnQ/d9IikH1b1/IUrnlrcacy1157Ko4NPVqkUstZT1dd7vf5E5dHBJwOOrzPW+3G+v5klJyqXSaKiLmkK0s7dg49ruGtj9LgTQcu3uDP42EZ7vihIAQCA5yhI0zsUpLVCSk7KtGWGvvfe+DO+9brQI/HPtnGnHTM+neb1NZbMRi7sOXPZ+/M9LwtrY5a+X2oeeqHda5Iurer5C1U8OcvR6pSkFKQUpNXdTrQFabjj2ejG4OMV6jq3snI9BSkAAEhBFKTpHQrStJeykzJtHXf6g/tea/mN12UeiX8+WXj7t7N9Gub1NZbMuL3eu9hus29Y1fPnVjytXH88qCCyl0O1pcCsqYLUbbRiuFCQhr/OnMWm/Xh17n7qGrRf06PHnQhYl1upWtXzleDjDgAAEIyCNL1DQZrWUnZSJkn64MWfL/566wuel3kk/jm8pdis/ddPXpf/GoWDb25+zhOv9f2v10Vhbc4/Fzzw5U1tMjtIOq8q59CteLIXSlbRFK40dRZGztLJXpDaSyz76+zrd97yvHJ94PedBZhbAWt/D26viXaZ8n3Bt2w7i1779p3LxqMgjXYEqfO42cvAOBR1Prn/mSggIQrScPFV9ToNdS5nlpwI+33rHEca+em8jilIAQBAyqAgTe9QkKYt56RMnypFJmWyvD/hzC+O7VzgeZlH4p+jO+ebLePOPKhqjNBLZ4UlHX3T3hpzxOuSsDZn6pvPfJP/WLOx8v9SKWahiqdoirZYC1K3WK91FqDOW6PD3S7tLBDDrcutkA21TPm+467bdr7naN5fogtSt+dohlt3uhWk1rF0u43eeS5HjzvhWpiGKsat65+CFAAApAwK0vQOBWnack7K1EQpMCmT3Y7JF+z0usgjicuW58//RP5r8nyvr7VkM3xBtw3Lyxd4XhLW5iwrn28endF+q6pY4ocqntwKN2f5VJWC1G39zhGiViFlL7GsbbgVW84C0Vq3fX/t5Vi0y9hvwba2Zd++2/uL5ZEE4Z5pGa4AditIrX23j550+1p1C9L+/fubeOjfv39CCtJw5aX9OEbzmALnOqM9XxSkAADAc8lekCrMb9CL55ZFtY7SdbsrXxPNtkrX7Y56nXmNm3p+jChIa52UnJTJadtYeV7ikcRl29jKZ+hd5fGllnT+ueiBnV4XhLU9mz/ZZIbN6bFH/uc4xzybfaTiKdwIy6rcYh9qRJ+9ILUKvWi/Fq5ADFWaRbNMqP12jiy0lrOPtI2lgKtuQRpuNKzb/idZQVqsKP5MRUEKAAAQAwrSqhWkeY2bUpC684U4X8WiII2HSJMyXacknZTJiYI0vWMrSHM8vtSSSuHcDpeMWNDzC68LQrLZDC3pdviaRr+7Q9KFsZ7HaGcHt4/4tAq3ZC1Iw5VZ9tGUkZZxjigNVZxFegZqqFu443WLvduzVGtrQcot9gAAAEqdgrQ664h3QXpxnSsql6UgDeITBWmiRDMp04882reYUZCmd2wFaZbHl1pSKZzT/pZnlg/c73U5SDabp1579FDTbg17S7om1vPoVjw5i6FYC6NkKEjdyrNw5ZbbMrGOIE2GgjSBRZ1PiXkGabHiUJBGmqTJ+UgH+3Xqdt04Jw2jIAUAACklXQrSnv2LAv7w2Kp9j4gFqb3oHP7czKgKUus11j8pSIP4VI0/zCOklJ+UyYmCNL1jK0jzvL3SksvAOe17TXnz6W+8LgfJZvNC2VNH2w67bbSqUOK7FU/2MsleNoUaQWov5uxfC/eMznDPIK1uQWoVmPYiyznCMJplYn0GqVcFqdu5cTtO6VyQOq8z+3u2Hy976W8/v86S1Pq6fUQqBSkAAEgZqVKQOmMvQFu17+G6jFVeuhWk9nLUmUgFafHcsspClYI0iC/EcS0WBWl1pPykTE4UpOkdClJ3Q0q7Tp63abrn5SDZbOZunGoeeP7ehZJyJZ0Vy3kMVTyFmr3dXspFurU7mlnsrcIqngVpuGdyxrJMuOPgNou9lwVpqMcFxHEWe8tFCvyFppVwBanb8s5cVJXr1Jlw16Pb7ffhrk3nuaMgBQAAKSPVC1J7+Wk9k9T5NWdBWjy3LKgMjXYEqRUK0pB8Ic5ZsShIqyotJmVyoiBN76RxQfq4pO6q4mfZY4vun7dy++KI5Z0ijCBbumGJ5wVjNPsf6vtLNywJWMb53zWRFdsXmv7TOq2R/zm558ZyHsMVT27lkXMZ52z3kUZY2tdpL5oScYt9uOIzlmWs0aahlkmGgtTtXDgfkVAbCtJQZXGoxy84j2e4YpmCFAAApIxUKUhjLSqtUaU9+xcFFaTWa+yjUO3bKl23O6BEVcUt9RSkUfHJ/S/0xaIgrYq0mZTJiYI0vZPGBelFkr6Q/+dxg6T+8v+SIipDS7tuX71jKQVpEhSkq3YsMQNf6fSR/KPxz4/lIoileCLpk3QuSJM5FKQAAKBG1OaC1PkaaxkK0mrxyf0v9MWiII1VWk3K5ERBmt5J44JUknoo+DNul6RRilCWDintunfth6ujLhiTvQiNV0HqRcp2rTKFr3b6TP5rtE4sF0C6FE/E86IumoK0ytLlOqUgBQAANSLVC1JusacgTVP/T2k2KZMTBWl6J80LUkmardAjPEOWpYNLOn+x7qO1cStIreWmlk4JOcJ01NiioH10rvfv+fcEfH/U2KKg1zdq0ihguS37NpveA3oH7INzv5zbDlWQuhWm0by3Lfs2m0vqXBLwvWiP3bqPykzhnE6H5L9Gr4zl5KdL8UQ8L+ooSL057gAAAMFSvSBdt6dmJ2miII3IF+K4FouCNBbFOnXs0mJSJicK0vROLShIz9GpW+3DJaAsHTSv4zfv7lkfdUHqlr/n3xNxuUvqXBJUPLp9360ctdJ7QO+QBau1jlDrjGW/IhWk4fbduQ/2RCpI39mz3hTO7fCt/Ndo/VhOfroUT8Tzos5IMj6frzK2a7ja20mX65SCFAAA1Ih0KEjX7TGmZ//Av8D17F9U+T23gtRZkjKCNG58cv/LarEoSKOVlpMyOVGQpndqQUEqSW0VuSANKEsfKs7fHcst6m5xK0itMtM+2tI5+jPcre720tF5i7y9IF26YUnANqaWTglbbtpHotq/FktBGuq92f/bKkOd+xrpOA+d032PTj3POWpdi5YZUjujBBSkIVLt7Xh9rJL4uAMAAARL9oKUUJDWMmk7KZMTBWl6J00K0u/I/yzgGyS1lzRC0suSNko6rNjK0Xck1Y11BGm0t9jbl3N+zbkv9rI01OhQK1NLpwSVrOHKTOc23W7j7z2gd0wFaaj3Zu2XvTCO5dg5RpDmVPEaySO1NvHgq0ixS+K5nbw0CgAAQGJQkKZ3KEhTinNSprclXaFTfylI6UmZnChI0zspVJCeJulCSY3kH739mPzPF90s6YhiK0FD5XH5b8lP2DNIo3mtc7/sJWOohCtI3W6pT0RBGuq9hRodG+2xczyDtEEVr588UmsTT4neTrj1p1oAAAASg4I0vUNBmjLcJmWqp8C/FKT0pExOFKTpnSQsSH8t6Y/yfwY+JmmupPdUveLzsKT9Yb7/oRzvP96z2FvLxfLaRk0aGSnwVnfnLfb2VKcgtW6Pt38t1lvsQ703t1vs7ZNGRTp2ZbtW2mexz6ridZVHam3iKdHbCbf+VAsAAEBiUJCmdyhIU0axTh2rtJyUyYmCNL3jQUF6uqTfSmos/3N7H5dUIul9BT62ItYclP+XFzPlL1Z7SLpd/p/PepIOhXjdbFWMGrUbWtp1++odS6MuSEPFmjHe+u9IoyzdYhWIVmHqjFWAVqcgDbXOeBSkW/YFT9LknNE+3DFetWOJGfhKp4/kv0arOoIUiIeLwgQAAAA1gYI0vUNBmhIiTcp0rdJgUianaAtSRSiKdr9d6rrszDHDg7727PCHKv89mm3vfrs0puXTMT3atwp5vD0oSM+Q/5cFf5LUTdJTkuZL2i7pWKRrJUy+kPSWpOmSRkrqLqmF/KVZhgJ/WWFPP5d1HZTUNNQbGLmo17yVHyyusYLUXnDa4ywPnTPZuz2ntCoFqXPb1vLxKkidxyrU69yyvHyh8U3vtLriXFb1GaQAAAAA0kFGds5/MrNzzX3d+npe5pH4p1XbriYzO9f8rs7l34iCNBlFMynTmR7tW0LFuyB1Fnll84qDvjZn4qikKkibNs6rLHKTcd3O4xfLsahGQXqGpDqSbpHUU9K/JC2UtFPS8UjXQ5h8LmmtpKmShsv/i4i/ScqVlKnob++0Z4FjG29KuiTcmxtS2nXyvE3TI5Z3JHLspeqosUVmy77gme7DZe7GqeaB5+9dqFOftQAAAABqq4ys63dnZOeazOxcM3t1ueeFHolfZq8uN5nZuSYjK8dc+Ovffi0K0mRTqyZlcoq1II00crFp4zwjKaAUdPtasiSR+xavdVvHvmxesTF715kr6lwc9XojFKTfU2JK0M8kvSFpiqShkrpI+qv85VemQo8EjSY58v/y4jL5n2f6C/lHnpqKf/aIeNFLGjAv//4pb/7rG6/LxXSJ8xZ7K86Z7d3yQtlTR9sMveWpivNbP5rzBwAAACBNZWZd/2FmVo7JzM41zVq2MSvLD3le7JHqZ2X5IdOsZRuTmZ1rrv59PfOri35zWBSkyaTWTcrkFM+C1CoErVxR52LXr4UaEepc1tpWqOWdIyudhaH19bJ5xa7rddu3aF4XzfZDrTvWWO/d/npru9EUpO8+LSPpXkkPS7pf0jOSXpN/4qIT9n2MMZ9KWiPpBUmDJXWS1ExStvwlV0PFtwT9ofyjWp2aVuzPO4rhc61wTvtbnlk+cL/XxWI6xXmNOGe1D5WnXnv0UNNuDXtXnPsroz2HAAAAANJQZtb1H2Zm55qb/tS0oiRtzUjSFM/s1eWmWcvWJjM71+Td+CdT57Ir7QVp2wRfUj5RkEajWKeOTa2YlMkpHrfY92jfqtoFqXM5+/fdlg9123mRr2fEfbbKxkgFaajXRbP9eBWkVS2rK4/rdVUuQI2kTyStlDRRUqH8z+m9Tf4CtLol6PWSrpF0qfyfRz9X6BI0nGL5P+9i4nu13W9HLOjxpdelItlshpYUHP6fc/7fLFX9OiWEkFTJp5LelfSypL7iucsAAASzCtIHC58yLe7JN5kVt9sPHDWB0aQplpXlh8zAURMqz2Hzu9qbe9r/w16QviTpjgRfUj65/8GsWBSkllo5KZP8575hRbq//Ej8ClJ7MRjuFntn4WkfqeksIIt8PYOWdxtVad9H539bpaXbdtz2N9Lrot1+Im7ft9ZpP+bh0uXWsH9ROylpt6Tlkp6XNFBSe0m3yn8rfDxL0Asl/UzSD+Sf6T5eLqrqC4sW997hdTlY27Np70YzfO4/9ijwl1WEEFKb8m9JBQIAAH5WQTqgaLyZU7bLFPQdXFmwNWjUhKI0BWIVow0bNak8dwV9B5tnZyw1rTs9YC9IJ8k/M3Mi+eT+h7BiUZBK6TkpU578ZdZfJPWvyCuSllXE7Xr40ItnkDoLz5ljhhspdOkXavlQsZ7V6bbPzq+FK0hDvS7a7UcqSJ238EcaZWqtL5bRqEPbyMg/YmWDpPEV18V98j93NFNSlqpXgl4n6ffyP8v0Qp0qQb8bxTXrqeGl3TYsL1/geUlYm7P0/VLTf1aHbaIgJYSQMUrPX8wDABAbe0G6bo8xkxdtMCPHvWRati2oLNsaNmpiuvUdwq33SZYpi9abbn2HBBSjLdsWmJHjXjLPzlgaUJD++uJLDksaLQpSL6XTpExNdWqCmqokrzYUpOHeR7iCNNTrvChI7bf0xzLi1DZJ099VvRK0rqT/k/QrSedJ+h+lQAkaTuHcjr7pbz33tdclYW3OlLXPfNt+5F/GyH+d5cp/jeURQkiappX8zwMfIf8jbJx/hhguAABqO2dBWvbvY2bakk3m2RlLTf+iCQFFaWZ2rmlyWwvz4KAnKUs9yuzV5ebBQU+aJre1CDgvLdsWmP5FEyqLUXtBesXV15hLL7/qoChIvZSOkzJdJGm2Yi9HZ0vezGIfyy32Pdq3ivoW+1D7HO+CNNrtx+sWe3shG81zR0MUpHcq9F/YGso/uZJbCXpata/QJDXw1XbXP/5a3/96XRLW5jw2/4GDN96b0Un+6zBL/s+zPEIIqSUplHRQgX8++4sAAKjNnAWplZK3Pq4s2R4b/7LJ79HP/OHmW4LK0m59h5jRk0u5DT9BWVl+yIyeXGq69R0SVIr+4eZbTH6PfgEjRp+dsdQ8N3NZ5b//49GRJjM711x2xdUUpN6aplPHIt0mZcqTf2byaMrRL1RxDcRjkiZFGDUZqSC1L+OMvZCMZnl7ael8TbiC1P7aWF8Xavtu665KQh3zaIpXW0H6V/kLqKvlv84vkPRTSd9XGpegkRTOzT++6ZONnheFtTEb975jBs7tcEynPn+vFQUpIaT2pZmkz3Xq/+/bBQBAbRaqIF23x5g1O46YWSu2BZRvQ0ZPNq079zI3OMrSzOxc06xlG9Ot7xAzcswMRphWMbNXl5uRY2aYbn2HmGYt2wQd4xtuvsW07tzLDBk9OWi0qD2zVmwza3YcMQOKxlOQeq+HAo9FOk7KdI78E+1EKjR7WC9IloLUrXS0yshQyztnkm/aOC/mEaT20auxFKTRbN9t3bHGeRu+PTEWpHlxucLSzPCF3acue3++52VhbcyS90vMQy+0e03+a7OB/COXAaA2+H+Sfiz/XUtXKvjP7S092zMAALwWriC1smzrgaCi9NkZS83IcS+Zgr6DTfO72gcVeVbualdgHhz0pBk9uZTS1KUMHT251Dw46ElzV7uCkMew+V3tTUHfwUEjRUMVo8u2HqjcBgWp59JxUiY3uZI2KnyZ+Y79BdEWpCQ1Q0Ea3oC5He4Ys3LIAa/LwtqYZ5YVHmrxQKNH5b82cyT90NurAQA88zNJZTr1Z7VJ8t/lAQBA7RNNQWplzY4jZk7ZLjNhzuuu5Vz/ogmmoO9g07JtgWl8a/Owpeld7QrMwFETzMgxM8zUxRvStjydvbrcTF28wYwcM8MMHDWh8r2HOjaNb21uWrYtMAV9B7s+U9QtE+a8buaU7TJrdhwJ2j4FqafcJmW6SqdKo1SalCmUH0l6RpFHjlplWSUK0vQOBWlkhXPzj2/a+67nhWFtisvt9Q09vQgAwHstderPauslZXi7OwAAeCSWgtSeleWHzCurt4csS+2laX6Pfqb5Xe1db8t3psltLSpHnQ4cNcEMHDXBjJ5caqYu3pBURaq1P1MXbzCjJ5dW7qs1GtT5vFC33HDzLab5Xe1Nfo9+UZeh9lL0ldXbIz77lYLUM26TMmUq8NlPqTYpk1MLSXsVfK5fc/lasfPFFKTpHQrSyIYt6Da9ZNP0k16XhrUp8zZNO9lrwt8X6NS1eZmnFwEAeC9bp/689h/5PxvP8XKHAADwRFULUufI0pK3PjbTlmyKqtzrXzTBPDT0aZPfo59p3bmXubX53VGVp25p1rJN5ahMt1jFZTQJtx6354FGkxtuvsXc2vxu07pzL5Pfo595aOjTMZehVqYt2WRK3vrYdaQoBWnSSedJmS6UtFDB5/gDSbdK6uD4+pdy+YM2BWl6h4I0Mt/c/JzRSx/Z73VpWJsyavFDBxv9/drOOnV7/XneXgUA4LkLdOrPbF/J//n4ay93CAAAT8SjIHVm2dYDpuStj82sFdvMxJI3YyoBR457qfJW/fwe/Ux+j36mZdsCc2vzu82tze8Oe+t+Tabxrc0r96ll24LKfbVujY/meaHhMrHkTTNrxTZT8tbHAc8UjTUUpJ5I10mZzpB0v/x/eLa/v2/kP++Xy//+7lTw+w9CQZreoSCNzrD5BTvX7FzmeXFYG7J6x1JTOLvzbnF7PQA4OR+JdLWnewMAgBcSUZCGKk3nlO0y05ZsinhbfiwZMnqy6V80IWSs4jKahFtPpFnjq5MJc14305ZsMnPKdlWrDKUgTQrOSZmKlR6TMoWahGmN/M+qylbg4wOs768ItUIK0vQOBWl0Bsxp/7dnlhd+6XV5WBvy9NIBh/72j7yHdeq6rOPpyQeA5OEsSLM93RsAALxQUwVpqNL0tU2fmTllu8wrq7ebyYs2xLU8TaZMmPO6mbxog3ll9XYzp2yXeW3TZ3EvQylIPec2KdM1Su1JmX4kaaykkwo8n/+V1F6Bk05ZqW9brm6oFVOQpncoSKM3ZF6XPWW7VnleIKZz3ti1whS+2nmfTl2TzF4PAKc4C9IcT/cGAAAveFmQhsuaHUcqb9WfU7bLzCnbZWat2GYmL9pgJi/aEPOt+4nKxJI3K/dp1optlftq3Rofy/NCKUhTWjpOyuQ2CdNJSeMkXSn/qNI8BRbAF0o6TdI7kh4Pt3IK0vQOBWn0Bsxtf9fYFcO+8rpETOc8u2zQwTv63jRAgb/IAQD4OQvSPC93BgBQob3ad8xXvgmX9mo/varrz1d+WcV6RlR3XyPs5x+ruM6LK15/cXX3LxrJWpDGmpXlh8yyrQdCxiouo0m49USaNT7ZQkFaYyJNypRKt3KGm4TpJkn1FDxq9Pfyl8SWYkWYAZWCNL1DQRqbYfN6vL32wzWeF4npmLJdq8zAWZ03K3D06E+9PN8AkGQoSAEgGaVRQRrzNvKV/0fbaylICQVpaohmUqbTPNq3WESahOkK+Sc1ybPlekm/cFnXRZE2RkGa3qEgjY2vpEODUYv68izSBKRo4YMHb2yT2VWnrkeerQcAgShIASAVVHdEZiKF2rf2aj+9KkUnBSmhIE056TIpUz3FNglTQ0n/J3+pWiUUpOkdCtLYDXyl86LSLTNPeF0oplNKNs441u/F+1YqcPToz7w8zwCQhChIASAVuJWQ9lGgzlGljpLR+l5Ht9e6LH9xfgwjQMOVt9Z27KNdw+2b2/fylV/msq1qj6K1oyBN71CQJlQ6TMr0I0lPSDqh2CZhqvbkJhSk6R0K0tj5lrU+q3Bex2Nel4rplIFzOpw46/unN9Kpa/EaT08yACQnClIASAURCtKAotGl4LTnYsdr3QrSmJ4lGqEgtcrbsor/Drtv4QrSfOXvDfGogY7O7caKgjS9Q0GaMOkwKVN1JmGqNgrS9A4FadU8MjO/x5S1zx70ulhMh0wpe+ZguyF/GSlmrgeASChIASAVRCpIHctapeQI5+ttIzVDFqSO0Z4RR5GGK0htz1LdG8O+Bd1ib/uafTRp0OjUqqIgTe9QkCZMKk/KFG4SphsV3SRM1UZBmt6hIK26AbM6lS3Y+vJxrwvGVM78LS8df2Ra/psKfCzIFV6eVwBIYhSkAJAKwhWk4QpCx3NAoypI823P/XQuF+2+2bYfUJBGuW9hn0HqnMCKgpRQkHoiVSdlijQJ02WKfhKmaqMgTe9QkFbLaYPmdvp6/e63PC8aUzFvf/ymGTS301cK/kXPdzw7owCQ3ChIASAVRBhBOsKxbMjb5aMpSB3riljCxniLfTT7FqqsDfU6ClJCQVqzUnVSplCTMC1RgiZhimTH5At2eV3ikcRlS/H5e8Vfsqrs9t6NbhhR2pNb7auQYaXdD9X/06X3KPDW+l96eT4BIMlRkAJAKoixIN1rLyXtr/egIA14fZT7FlSQ2kebxrJv0aIgTe9QkMZVKk7K5NkkTJG8P+HML47tXOB5kUfin6M755st4848KP6SVS2d/9m8f/GqUZ95XTimUopXjfqste/WYQr8Zc9VXp5HAEgBFKQAkApiLEhDFo41WZDaRo/anyUazb5d7PK6oP2oTSNI5V62GUmmeG5ZVOsoXbe78jXRbKt03e6wy7Vq3yNgPyItT0GaFgVpKk7K1FQeTsIUyY4Xfrno660TPS/zSPxzeMsE88bon5SJv2RVW7enW02atObJA14Xj6mQSWueOND1qTuLFfi5luvd2QOAlEFBCgCpIMaCNGh2+xoqSENlhHN9EfbNOdN9maNsrXW32CvJClJnORrNeilI06IgTaVJmS6U9IqCz8EWhZ6E6WrFeRKmSN6bcHrvfYtbfuN1mUfin08W3X70hV56quLayqrJ6yoNfb/3861LXnp74iGvC8hkzqy3JxzqNf7v8xT4uZYj6WzvTh0ApAwKUgBIBbEUpI7lTcWkRiMct7rXVEHqdtt92H2rWMZeiO517LOpuOX+j/bvV0eqFKTVWUc8C1I5ytmL61xhJJnhz830/FhRkCZMHwW+l/uVnJMynabwkzD9n2pwEqZIzHM6471xpx33uswj8c+2sTp5+um6QadKKlTPLx9+sf2CCUseP+J1EZmMmfrms1/3nXTffAV+tl0njz7bACAFUZACAJAuBWnP/kUBhVyr9j0iFqRWuamKglMRClJrPRfXuaLya9aIUgrSSj6lV0Gao8Dnd05Wck7KVE/SW3KfhKmePJiEKRrvP/+Df3y6rMN+rws9Er98suS+Lyb3PdsaPZonnv8YrTz5fy67S+ov6XlJyyRtkP9nuazPpLbzZ617nombj5Mx7AAAIABJREFUbJm5bvyhB59v7Rw5er2kX9f4GQSA1EVBCgBAqhSkztgL0FC3vec1bhqyILWXo87E8kzRqryGgjRlpMKkTKEmYdor6S5JVyj4dvoM1cAkTNF6+7lzZ+1b2vaI18UeqX72Lbn3yJLHflyqwCL+fC+vryRxjvzHon9FXlFg+RlN6kqq82Bxm9JJa57c73UxmQyZtOaJA72fv9fttvr/q6kTCwBpgoIUAIBUL0jt5ad127vza86CtHhuWVCxGc0IUmfyGjcNKmuTLRSkVeY2KVOWAv8i7nXxE24SpksVehKm79T4nob3nZVP/PS1T5d3Oux1wUeqUY4uzT+4fNR5CxVcyMMvT9KHir4Qtedx23ouLXiq5cSJqx6v1bPbT1w16rMuj9/xvIInZEqm50EDQKqgIAUAIFUK0lDft4pNa7SoFWtUac/+RUEFqfUaZ7FpLVO6bndAiSrHbfX2ctT59WQLBWmVuU3K1FCn/tB4qVc7psiTMNkfAWClxidhitH50/ud/eS2sd89vmf+377+emuxObpzvuelHwmdozvnm8NbJpg98/96eNtYnXTcVp8n/23OXv8SIRn5JH2h6MvRL+QfgWpXp/WgvwweXtrz4PqP3/K8rKzJvP3xm2ZYafdD9/r+NESMHAWAeKEgBQCgNhekztdYy0QqSO239Ht9fChIEyJZJ2UKNQnTV5J6y30SpmxJ59X4nlZNnbNOV96kBzT69ad+/PqWsWce3Pz8+Xu3jZUhyZUtxefv3TL2zINrnvrx6y/00lO2CZmsNBBlVTjnyP+ZGE1B2jrEOupc37zuXYVzOh5euOWVY14XlzWRBVtePlY4p8NX9f906T0KLuO53gCg6ihIAQBI9YK0pm+xr8qt+BSkKVWQJuukTOEmYbpKoSdh+m6N72n1XKbgRwOQ1EqOvB1hnSrOkDRG/sdihCpH34mwjl9Luu7R6flvT1v7XFpP3jTljWcOPTqtw1sKvt6uU/L/fwUAkh0FKQAAqV6QrttTs5M0hXoNs9hX8oU4RsVK/oI0GSdl+r5SfBKmKjhf/oKXojS1Yp0vbquPrJ785Wek0aN5UazrF5Jy2g1p+s/CuR2Ol26eddLrMjOeKdk84+TAuR1OtB3c9DG5l/G/iON5AYDaioIUAIB0KEjX7TGmZ/+igL9Y9uxfVPk9t4LUWZJGMzLUedu9PRSklXwhjlGxkrsgTcZJmZoqeFKXE5KelPskTLnyl7zJNglTVZ0v6Ur5SxDnuSDJkSz5z8+VohiNxpmSihT8C499cv/MjNbZknK//6Ozbnx0aoc3nlj88OG1H672vNysTso+XGVGLe77Vb+p+WvO+v7pjRR43Vm/QDk7DucEAEBBCgBA8hekhIK0hiTTpEwXKvQkTDkKPQmTF7f+A4hOrqRyBf5MfyWpu6RRjq+7TcwUjasl5TS+L6vb4Fe7l49dMfTIG7tWel52xpI3dq0wY1YMPTL41e7lN7bJ7Cr3UaNXV/00AECtdY6k2XIvP8MVpHkVr6vK/5cAAEgdGVm5uzKzc02/Ec95XuaR+KffiOdMZnauufSKul+KgjSUZJmU6TSFnoTpAaX+JExAbfQjSc8q+FmjC+V/5m49+T+b7d/rU43t/a8qPifu6nfzoMJXO//n6aUDD67ZsdTz8jNcVu9Yav61xHeo8NVOn7bq12SgQo9a/mU1jg0A1HY++f8/s0yBJahbQZpXsZypeB0AAOktMytnS2Z2rmnbpbfnZR6Jf1p3ut9kZuea39W5/HNRkLpJlkmZIk3C5LzFvKGk3yr1JmECapOm8j8v2P4z/Zn8zw++WKd+4dHD9v1/x2G735H/sQe5kvKa97rhkQEvd/j4nwse3D9v0zSzce87nheiW/ZtNhv3vmPmbpxq/rngwf0DXu740d/+kfew3EvR3Ir3ky6PDwEAr5wj/10K1v9zrKLU/v+pHpI22P67qnc1AACQWjLqX/96ZnauyczONfPe/NDzQo/EL3PKdpjM7FyTmZVjLvrNxXtFQeqUDJMyVXUSpv9J8H4BqLqfyv0xGZPkn3k+U4E/0wW2ZfLiuB8/kv+XL7mS8m5sk9m11/N3Lyyck3/82WWDvlr6fmmNl6Ub975jlr5XYp5ZOvDQwLkdjvUaf/f8Rq0zOsu9GG1Qsf+pPOkcACQbn8JPEOiMz4udBACgxmVk5Sy3CtLb78k3qz/42vNij1Q/q7d/ZZrf1c5kZueaq36fYX514W92y1+Q/i3Bl5RPqVGQJsOkTLV9EiYgHbWSf5So8xcejSX9TsGPybhK0lkVy81O0D79XP7HceRY2729z42+fi+2X1Y4N//4qEV9Dk1787njy8rnm82fbIprIbpp70az9P1SM/XNZ48XLXzwq8K5HY49PLn90hYPNHpU7qVoXsV+Zkv6WYKOBwDUZs5RpOHC6FEAQO1hFaR//HMzk5mda5rf1c6UvvWR5wUfqXrmrv3Q/LVVG5OZnWtuuOkWU+eyK+0F6S0JvqR8So2C1MtJmX4pJmEC0s2F8j9X1P4zfVL+54/+Wv7CL8+W6xVYAH6hxH8+/lRS/YptV+5Lo79ndO70ePMJA2Z1LB82r+feYSXdvx79Wv/DU9b+69jcjVPNivKFZtWOJaZs1yqz7qMy886e9WbLvs3mnT3rzbqPykzZrpVm1Y4lZnn5QjN341Qzee3oY6Nf6394WEn3r4fN67l3wEud3u886m/jb7w3o5NCl6LWMcms2E8AQOL4xOhRAAACWQVpn0FPmia33W6s0aQ9Hh5uXtv8medlH4k+izf9x3TrO6TyHDZpeoe5J7+nvSB9SvG9fdONT8lfkHo1KdNpkjop+Lf2TMIEpK7T5J+J3jm5Wrn8P8Nuj8moI+kMx3qaJn5XK/2wYh8aKniket7v/3BJy6bdGvZuO+y20X0mtlnsm9759YGvdP648NXO/y2c0+lQ4dwO33Yd1fI/hXM7fFs4p9OhwjmdPhv4SuePfdM7v95nYpvFbYbe8lTTbg17X9Xod3c61+2SnIr9qCNupQeAmhLNKFJGjwIAaherIB1QNN7MXfuhad2xV2XBlpmda/oOGW2Wbt3veflHQmfZls9NrwFFAeetTZfe5tkZS03rTg/YC9IB8v9lNJF8Su6CtIm8mZQp1CRMc8UkTECqukTBP9dHJRXK/4iOgJGa8v/C48c1v5thnSfpcvk/c6yyMi9MekjarlOf6bHGKmUbSrpM/AIIALziE6NHAQA4xV6QrttjzORFG8zg0ZNNi7vbV5Zt1+X+wRT0GWxefeMDz8tAciqvvvGB6fpgocnOuaHyXDW/u70Z8vRU8+yMpQEF6W8uvuQj+f9iW5sL0gsU+NvympiUKdQkTJ/IP2KMSZiA1HOGpL6Sjijw53qj/L9wuUrBpeDvJJ1e43samx/K/6iAevL/0iZXp8rMWyTN1Kn32kPRl6G5Feu7tmL9jBQFAO+FG0V6UIweBQDUNs6CtOzfx8yk0rfMszOWmiFPTzV3tu4UMDLxthb3mL5D/2UmL1jneUFYGzN5wTrTd8hoc2vzuwLOS8u2XcywZ6ZXFqP2gvTyK+uayy6/aodqd0HqxaRM4SZh+p1sE6aISZiAVFFP/iLU/nN9RP7C9CIF37Keqfj/4qWmnCXpXEntJO1W4HsukP+XSlkV/8yRfwR+fUlXyv8c5/MrXn9WTe84ACAqPjF6FAAAP2dBauXlVeWVJduIsbPMXfd1M9m5NwSUcg3+cLPp/IDPjHt5uXl95xHPy8N0zJodX5sxs5aazg/4TO4f/hhw/LNzbzB3t+9uRo59OagYtdL9oaEmMzvXXH5F3Q9UuwtS+6RMJyT9WYmblCnUJExvK/QkTFeKSZiAZPZ9SUUKHg2+Uv6R4M6f6wbyT86UiOcZ15Rz5P5ZZpT451kDABLPbRQpzx4FANROoQrSdXuMWVl+yExbssk8N3NZZeHWd+jTpmWbLgG3dVu5t+M/TOETxWZSyVqz+oPDnpeLqZjVHxw2k0rWmoFPFJt7O/4j6Bhn595gWrbpYvoOfTpkKfrczGVm2pJNZtX2r8yAovEUpMGTMj2swNGb8ZqUKdIkTJco+Pl+10n6SRy2DSBxchU8GvwLSfdJ+pX8ZWieLfXkL1RTWQ9JBxT62XT85RkA0oNPjB4FACB8QWpl1favzIxlWwJKuH9NXWwGPj7RdOldaJrf1d7k3nBzUJl3W/O7TbeHhpjHi181s1//wLz58QnPC8hkypsfnzCvrtluRhXPNt36DjG3Nb876Bjm/uFm0+Lu+0yXBwtN4RMTzdPTFocsRp+dsdTMXL41oJymIK2xSZmuFJMwAenmR5KelXRSgT/XC+X/+a2n4FGjv1JqPybjIknLFH7iDuPVzgEA4s4+ipTRowCA2iuagtTKmh1HzKwV28yYl1a4lnPDnpluevlGmXs79TJ/anpHUNl3fYNGpmXbrqagz2DjKxpvnnqhxExe+LZZuH6v52VlIrNg/R4zecE68+QL84yvaLwp6DPItGzT2VzXoFHQMWrS9A5zb6depteAx12fKeqWMS+tMLNWvmfW7Ah+zEEtL0gjTcrUUNV/NuD3JQ2Vf+Zq+/v9SKEnYbpWTMIEJLumkvYq8Of6M0l3SbpYwaPB68r/rONU5lPkYpSCNPGuk9RG0iPynxN78rzZJQBpzidGjwIAartYClJ7Xtv0mZm5fGvIstTKwMcnmoI+Q8wd93YyjW6+NagQtOePt/zNtGrbxXTpXWj6/3OcGTFmuhk9udRMmL3KTHvtXTO3bKdZsvm/nhee6/YY89rmz8zcsp1m2mvvmgmzV5nRk0vNiOemm0cfG2u69C40Ldt0Nn/881/Dvt8/NL7V3Nm6k+nWd4gpfGJiVGWovRSduXyreW3TZ2H3sxYXpNFMyvSram7jjwo9CdNvFHoSJgDJK9QzhCfJ/0xR5+dITsVrUlmepA2Kvhz90IudrAX+Imm7wh97n1c7ByCtnSP/ZzujRwEAtVdVC1Irb+0+aZa/96WZ99ZHZtqSTWbc7NVhi71/TVlohj4z1Twycozp/vAw065bX3P73zuaPze70zRo1CRsoegcjXrTn5qav9z+d9OqbRdzd34P07pTL9OuoI/p0LOf6fyAzxT0GWR6PDzc3N//MdN74OOm79B/mX4jnjO+ovGm34jnTN+h/zK9Bz5u7u//mOnx8HBT0GeQ6fyAz+T36GfaFfQxrTv1Mnfn9zCt2nYxt7W4x9z0p6auoz5DpUGjJubPze40d9zbydzX7SHT/eFh5pGRY8zQp6eZ0VMXxVSIjpu92kxbssnMe+sjs/y9L81bu09GdX5qcUGayEmZfin/rfrO9/i2/CN/mIQJSE33KfgZwnslNZb0OwWPGr1K+v/snXmYFNW5/98Ylyy/3JvFuAUCEREQZJml6YZuHDQat6hZRBCMiALCAAIDCAKZYhFh2JUZdmj2HVmHYZFB1gHGRCMNKptGQW/cNaLR6Pv7o6dmqquru6t6qvtUnf5+nuf73DgzXX36vHW6b384VS9dmv5h2sZPiWgqmRejavYIGKvsFJC5uVcSHWjmoBV5xQOXKQiSqZk5aEVeqhaqGdy6Bp/u/tw00WNwUh2tkOP153m8fiXTkuP154meewAAsJXaClKj7D/1BW9/+T1+fv9JXrKt0pIInLV6N0+YvYYLJ83jwaOncp8nx/Jj/Z7iTt368h86duXb772fb/qteZGaitz02zv49nvv5z907MqduvXlx/o9xX2eHMuDR0/lv06ezxNmr7H8mvVZsq2Sn99/kre//F6tGl5lqCBNVVOmWE2YPiGifApfdqtv1uIlNGEyTWg1XXpiDg16Y/HPt76+4NKPTy2re/rEXGLEWTm1rO7p1xdc+vEbi3++9cRcKuDZdInoc6eW1KPwfUW16/o7Ct9/9DdE5KPIdd2WiK5I/zBTSkuqeX1diWgORd97FYI0NTQn83J6GoU/R2Pu8ioeuEwpKVjOiLsys2CF8DHIkuKBy5SUrtgEYA3KUUcreLx+ReR3U2Hx+hXRcw8AALaSCkFqlD2vfcpbj77Nq8tDvHDToVoLxFmrd/O0RVv4mZKVrExdyCMmzuHh42fxsHHF/OTYZ3nw6Kk8sHAy9x9RxP2Gj+e+Q5/m/CGjuVdBeIdorwKF84eM5r5Dn+Z+w8dz/xFFPLBwMg8ePZWfHPssDxtXzMPHz+IRE+ewMnUhj5+5iqcv2mLLuPVZuOkQry4P8dajb/Oe1z61dd4zUJCmqilTvCZMjSh2E6ZkRGxG8sbinw15bf73//Pezgc+//L4Yv7mTBnz+UrEofnmTBlfOL6Y39vZ4fPX5l3035OLfjJQ9DmUBBcR0RNE9G+KXNdvEFF7Mr6HcGMi1wthMxyhmvn4hiLnZ4PAcclIBZkXpEFK8FmqyplVE0t555KDiAuyamIpo2Y2zOOkUkeINaxBOepoBVWQdn60DytT5kufzo/1hSAFAMhJugSpPkfe/pb3nfw3v3DsQy6tfIc3HDzNq3Yf46VlL/H8jQdSIiJFZf7GA7y07CVeXR7iDQdPc2nlO/zCsQ95/6kv+Mjb36Z0njNMkKaiKVOiJkxNKFqgoAmTRU4u/uXqD/bmfyRa+iHJ58N9+R+eXPKLRaLPJQsY/aPH10Q0hsI7SvX3EPYR0c/SP0whdKWaOfmKiJ6iyHlSRA1MQpqStdsbBMmkIN255CC/FXoXcUF2LT3EJQXLedfSQ8LH4uao8/h09+cS7rROJViDctTRCqogTff3aVEZNXUBe3wBbpmV65oaAQCAKUQJUjPZf+oClx//mLe//B5vOfpP3nDwNK998QSv2PkKLy49Klykzt94gBeXHuUVO1/htS+e4I2HzvCWo//k7S+/x+XHP+b9py4In8MMEqSpaMpk1ITpGyIaT8ZNmPxEdE2yLyBTObX06nXvv/j4BdGCD6l9/lXe7fM3lly5QvQ5lYBLKCxB9f/o8Q8K/+NGc4reDX49EV2c9pGK4XIiep9q5mUchefgZYIgTQUjyPgz8ysiWkfhz05t+hMEqXRRhdDW+Xv4wI4j/GrF68LH5Mao8zi089hgonWSSrAG5aijFTJVkDa9sWWQXFIjAAAwhZMFqdlUvPk17z/1Be858QnvDn3Eu459wLtefZ93/uNfvP3l97js7+/ytpfOcWnlO7z16Nu85chbvPnwm7yp4ixvPvwmbznyFm89+jaXVr7D2146x2V/f5e3v/we7/zHv3jXq+/zrmMf8O7QR7znxCe8/9QXXPHWN8Jfs+kPsMwRpHY2ZYrXhCmbYjdhcnOzFiGcXPLLp95/sff7osUeYqck7fnBG8GfDxJ9bsUgh8IiVLuuvySiYRR+bwpQ5Lr2kPVd525nFtXMzQkiak3huehIRJ8RBKndvETGn5njKfpzRpv6sQ4IOeO+qEJoXUkZrwqu4/07jggfkxvjFLGGNShHHa0AQer8GgEAgClkEKRInA+wzBCkdjVl+h6hCVPaCK2mS1+bf8l/RAs9xP6cmHfRN6w46v67PyaiKRR5f2Imor0U3jGq/0ePdhRuzuSk15AOconov1QzP50ock7UTut5QkYnH3XI+PPyAsWXo3kEQSpVIEjtnUfRYg1rUI46WgGC1Pk1AgAAU0CQyp0MEKR2NWW6kcKyRD/u5RS7CdO1lHkCxTaOz7v4yfd2dfpKtMxD7M+72zv8Z4NC40WfY1UY3SrjEyJ6jMISVP+PHjkUFqqZiLYx00YiakvRQu5lgiC1i7vJ+PPyONXMe4DCDf/q65Kwiz3kjHsCQWrvPIoWa1iDctTRChCkzq8RAACYAoJU7kguSO1oyhSvCdNtFLsJ04/sehGZyqmlV+68cHyJcJmH2J8vQkE+XPyLgxReo6L4XyJaTNHvRdspLJxyKHrX6K8pvJM8E+lKkfe/1IpjD9X8Y1BLgiC1iwfJ+PNyD9XM/VVWDwo5475AkNo7j6LFGtagHHW0AgSp82sEAACmgCCVOxILUjuaMqEJk0BeX3DZJ9+cKRMu8xD78/WZbRyad9lnFP5HChHcR0TnKXJtv09EnSl8qwzt/YnzKCz9fpj+YTqGWI2Z8qryc1EDk5yeZPx5uZVqdo9aBnLGfYEgtXceRYs1rEE56mgFCFLn1wgAAEwBQSp3JBaktWnKlKgJk7pDSpsmhCZMtnJ6Wd0zokUekrqEFtZ5l8K7EOuk8bS6moiep+i1vZjCYlT/jyj+qsdkOrEaM+URUTNho5KfQWT8ebmawnOflcxBk5UzRMSNGjYWLkesRp23mVPnRvy8UcPGTETCx2dFCMUSpOprISJ+9C/dI343c+rc6t8lU7/f3XIHb1ix1ZYa6KOvSarH4xSxZrQGD+/+W8x5qu38J6qN9rxQn7M2NUnmGG6soxXMCtK82++LeR7k3X6f8O+Nebffx8HNFYm/X0KQAgBkBYJU7kgqSJNtypRsE6aY93oDyXNiLgmXeEjqcmJudTOf5mk4nb5H4XuK6tf2m0R0OxE1puh/9GhO+EcPosSNmczcwxkkRyHF/7xsmsxBM1WQ6scuqyDVv85H/9I9aUGqPjZVgpSIeOSQUWkbj1PEmlVBSkR8ePffUrY+arOu7TpH3FhHK8ggSB/s3p+JCIIUAJDZQJDKHQkFabJNmdCEyWFAkModjSBN9XtOPYpe299ReFfkr4nIR5Fruy0RXZHiMbkJM42ZQGpQKP7nZaNkDmqHIN2wYisTEf/uljv4d7fcEbFjbOSQUYY7BNXHaKOXPlrRpwojvbzRHl//HInknFbG6QVpvPEl83qTGWsiIZRIkKr/V7+bT/t7s+PTvk71tZutoxkJp93ZaqYGdozHKWItniCNJbjVc1edB634Vl+vmfNNu8bUGiTaQaqf+0TniNEx9GPTi3HtuWv0XE6soxWsXmJfWvlOlBQNbq6IOtdLK9+p/r0qV1WRqf19g8bNIn5GRNygcbOI5xxQOCXi2BNmr4k6tppEshaCFAAgLRCkckcyQdqIrDdluozQhMmRQJDKHY0g9aboFLqIiJ4gon9T5Np+g4jaU/jS8DxdGhPRJSkajxvpSjXzFq8xE0gNCsX/vGyQzEHtFKT6aOWLGq38Mfr7WLLFSOpppZA28XYh6selyiStIE00PquvN9mx1laQqhJKvcxeHbc6FitzaSS/zNTRrCDVPseGFVsTHtuO8ThFrCUjSFXZabROzJ5v+sfqj2EkN40eE68mRseINTbt7SCM1pL2mE6soxVqK0jV/9ZHKzn1ElP9ndHP9Y/VSlVtBhROMTwGBCkAIGOBIJU7EgnSHxLRa5rjfUKJmzLdRmFZoh0HmjA5BAhSuaMRpHkpOH1uJKKjFLm2vyaiMRTeUapf214i+lkKxuFm0JhJPAql4PPSbkGqF4zq7kX1v2Pt5tM+Xn887Y4z9TljCaRGDRvHFXTqY9TdcqpwSXSJfbzxJXq9yY61toJUHademul/bnZ8Zi6f1tfNiiBVaxxrZ63+2LUdj1PEWm0usdfuIE0kV7X1NFpjiXaQ6s8bbQ3UmhnVRHsM7etSX4PR+tFK/ljP7bQ6WsGOHaT6UNUcqv+t3UGq/ky761T9mbpTVBWksXaUNmjczFCi4hJ7AEBGA0EqdyQSpPqmTA9Q7KZMsZowHSA0YXIMEKRyJ0WC9DIKS1D9jvCjZLy2byKi64no+zaOQRbQmEk8CjlckGp3d6nj00u7eJfRq3+vihp9gyHtc2ovyTZKPBGr3/E2c+rcmILUaHxWX2+yY62tINXL0kYNG1c30ElmLuMJyXjzpEZ9Pu3/NiNIjY6dzHicKtasClLt+aLddasXnfEeb2aN6c/rWI/RJpEgjXUMfe2N3i/0Y3NaHa1gpyDVXi5PBoJUKzAnzF4TJU0rz3GEEFX/JlbUy/QhSAEAgDJPkPYYWMg9ChTuUaAIH0s6Iokg1TdlGknGTZkSNWG6ltCEyTFAkMqdFAjSHIreEf4lEQ0jorpEFKDIte2h6FtugDBozOQMFHKZINXKDK3wMNrhpr4euwRprF2FRuNWf6Y+f6LxWX29yY7VDkGqvZyeKLwjz6ogVcenl19m50k7x/FEl9El9vEu97Y6HqeKNSuX2MeaM61INFNPtwjSeGNzWh2tYNcl9todnRRDkGrvS2qHIFWFKAQpAABQZgnSZTv+Xv1hcP0NLYSPJx2RQJAaNWXyUORl8ZdR4iZM2sfkEZowCQeCVO7YKEh/TERTKPJ9gCm83ptTZJO2PAoLvt8Q1nY80JjJGSgkiSBVZYj+8lmi6Mtw9QIl0WXhVsatFTra50s0PjuEcG1iRZDqZdmGFVtNX2KfSH4lmiezNTB6rJlj13Y8ThFrdgtSM481s8b0cjPerS/UOa/tJfZG9wSOd944qY5WqK0gVS+LV+8JanTpvJEg1d67VP2Z2Uvs9YEgBQAAyixBOntNefWHSLbvJuHjSUdcLkjNNGW6moybMJ2i2E2YWhGaMAkHglTu2CRIbyOiNylybX9CRN0pLEH1O8KzKSxUQWy6Us1cfkZozCQShSQRpPF2uOnFjxqjJk1mmsyYkXP65080PquvN9mx2iFIjV6f0f0czYxPK5N/d8sdpuoYqwaxooo2M8eu7XicItbsFqRm62lXkybt8+trYnQMK02aIEjD0QvSeLs81ccYCVLtz9UYNWnS/432b9W/0TZyQpMmAEDGkkmCdFPF2fAl9gML+YmRE3lTxVneVHFW+LhSGRcLUjNNmTpT7CZM9QlNmBwNBKncqaUgvZyIFlP0+8jzRHQdRe8Ib0fRTdpANPrGTH8lNGYSiUL2fF5GIEKQ6kXLyCGjqneiaXeeae8jGUuQ6GWLUZfrRJJFK9aMRJB+fMm83mTGapcg1TfyidXwJtH49LckMFtHs4JU/5hEx67teJwi1lIhSM2eb9o1lqhJUyxJqn1uo5oYHUOtC2lqFe8chiCNvgeKp5JrAAAgAElEQVSpVmIOKJxSvRNU3dEZS5BWnou8b2nluchL7NXoO9nrJah212qi3aYQpAAAackkQapm7xuf8+w15Tx7TTkvLj0qfDypjIsFabymTPcS0WaD50ATJhcBQSp3aiFI76NIicdV//1HImpAkUIvj8Lr/Ye1PiEzAzRmchYKOUiQpjJGXa3jdbHOtCQSpIi1eRQt1py4Bt0Up9TRClYFqV3RXmKvilRVdCaSnLX6fglBCgCQlUwUpOXHP64WpMt2/F34eFIZAYL0r1T7L3zxmjI9Q0Qf637/ARE9QsZNmFoTmjA5EghSuZOEIL2awjtE9e8diyksRvU7yP1VjwHmiNeYSb2XM0gvCmWIIH0rFN2FXI1+p1kmBoLU3nkULdacugbdEqfU0QqiBGnlueiu92rU+5mm5PslBCkAQFYyUZAeOP1l9eX1ZX9/V/h4UhkBgnQIGX/hW07mPkBjNWXqTESvxDjutYQmTK4DglTuWBCk3yOixyjyfsNM4XuP3kFEjSlybedReLcjdoRbI15jJtyeQAwKZZAgfSsUfSl2vO7ZmRQIUnvnUbRYc/IadEOcUkcriBSklec46r1V39Xe9u+XEKQAAFnJREGaSREgSPuS8Re+tZT4A9SoKVNrIppPxk2YbiU0YRJN0rtzIUjljklB2pDC3ei1a/tbCnet/zVFSry8qv++PNlzLoPpSjXzi8ZMzkGhDBOkSHwhBEFqzzyKFmtYg3LU0QqiBWnav19CkAIAZAWCVO4IEKSPkfEXvk0U/wPUqClTPyJ6W3ecryh8GX8seXKVvS8HJGADERVSEqIUglTuJBCkFxHRMCL6kiLX9xsUvpS+GUX/w0djIrrE6nkG0JjJwSgEQYqEIEjtnkfRYg1rUI46WgGC1Pk1AgAAU0CQyh0BgvRBMv7CV0bhD9BrYzxO35Rpn8ExDhDRDRS7CRPkSfrJo3BtPiaLohSCVO7EEaQ3EtFRilzbXxPRGCKqRzX3G1bjJaKfJXd6AkJjJiejEAQpEoIgtXseRYs1rEE56mgFCFLn1wgAAEyR6wu84PEFeETRbOFvtoj9GVE0mz2+ADdp2vINSo8gvYWMv/AdpfAH6HUGj9E3ZfpC998fE1E3CssTNGFyHnsoslamRKlZQUrG51N13nmp1PBv18yZEPWzWROeqv7fZp77nZdKLf29jGnWuEH1HNx3e15tBOllFJag+ttlHCWiHIr+h4+bKNyc6ftWT0hQDRozORuFUiRIZxas4JVFW3nnkoOIC7JqYimXFCzn+cpqnvRUMS+Zsk74mNyYlUWlPLNghXCxhjUoRx2toArSzt36sDJlvvTp/GgfCFIAgJx4fP41Hl+Au+UPES7zEPvTtVcBe3wBbnh905cpPYL012T8he8DCn+AXq/7e31TJm2+I6LNFN4dqm/C1I7CwhT3zhNPHkXXLqEotVuQ9u/+YMTPK7YEo362adFURwnS+27Pqxa5Tjz2fbfnRc13/+4PJiNIAxS+fF57rH8T0RNk/A8fHiL6SdJnJFBBYyZno1CqdpAOXM4lBYhrUlWvyfnzeECHkTzmsenix+TWDBS/8xBrUI46WsHj9Sser589vkDmxOuHIAUAyEdu6zaT1De6LUfeFC70UplNFWe5R4HCPQoUnjT/eeHjSf3rPV39AfbretfupvQIUqLo+wqqWUJEMyj8pVAhoueI6EKMvz1H4YZPPjJuwvTDNLwOYJ49ZFzHmKLUqiDV7hSNJ/O0UtDoZ05JKsdmx7H1crhiS5CJiJs1bmBFkN5O4fsPf6c7L/ZS+J6i2RT9Dx+/IfzDhx10pZr5RmMmZ6IQdpAiS7CD1K44Zech1qAcdbQCdpA6v0YAAGAK9Q3d4wtwh4d68P5TF4SLvVSlx8DC6i8gPQYWCh9PKrP/5L/5z50fZY8vwM1b5XKdOr8uo7AgbZuG0+plMv7SZyZfE9ECIupARH8kNGFyC3kUv65RotROQarf6discQPDn8XaEar/W/W5Yv29fmeqXkSqP1elov64RmMz8zgzzx/r2LWNOiazl9nP609MRO/pzoNPiKg7hSWoftdoNhH9uBbnIKgBjZncgUKp2kFasJx3LjnIr1a8zgd2HEEcnvUlZVxSsJwXFa3hGZNm8drFG4SPyY1ZNm09j+k+nbvd2WctEd2d7BqqLViDctTRCtp7kJaf+Jg3V5yVOgNGFnGLrFy+rmEj19QIAABMob6h33bXH9jjC/CfOz/KpUf/KVzwQZAmn82H3+Q/PvgIe3wBbn/r3dz4hhu1gjQrDafVWDL+0ocg1aLUjkvs1cu9ayNIjS4jV39v9Pd6OalmijIg4ZhVWZlIkMZ6nJnnT4UgnaIMqD5eop28fL6Sz/1tm9EYnyeiRhR9u4wAEdUhYCdozOQOFDJe80GySZDu33GEVwXXIQ5PcMJqLilYzsWFC7lw6Gh+btJM4WNyYyYPL+YBHUbyPf4OZUTUMdk1VFuwBuWooxW0gnRTxVmevbZc6jyS/yQ3adqc69St75oaAQCAKdQ39KFjn+U77ulQfV+R/sMn8K5j7wuXfXZm9ppy7jGwkLv1G85Pl6zgTRVneder8rzGna/+i/sNG1ddwzvue4Af6jFAK0j7kXGTJLtpTOJFHOLsvLnrGfsEqVYMxrvEPtZl4+p/awXkFGVA1N+r/62XjvpjqP+tSkuj5zEab6LHmX1+uy/f10pZrQiOl/zfV9fpAwrvBr+eIncx5lG4MdMPCNgJGjO5B4WM39uChB2kGRXsILUnTtl5iDUoRx2tgB2kzq8RAACYQvuGvvnwm9z18UERN2AeNm4G7z7+kXD5Z2c2VZzl2WvKefaact5UcVb4eGqb8tCHPGjUlIi6PZI/hGet3s1dew3WCtIHKH0fYG+ReAmHOC+fUFgKWN5Basc9SPXCc82cCUwUu/FQrL+PlYotwZhj1v8sniCN9Tizz59IkOov4Te7y1TtaK8+T7xU3YN0GYXfd7wUKUb9RHQ1Abv5PqExk5tQyHgtB8kmQfpW6F3EBdm19BCXFCzndSVlvCq4jvfvOCJ8TG6MOo+i712JNShHHa2g/T4t+nthOjJq6gLcgxQAICf6N/RlO/7OT89Yxvd36V4t29oEbua+Q5/mjYdOCX9DtiOyCNKNh05xnyfHsM/fvrpWf+7SnceVrOBZq3dHCNL69RtspPR9gHUk6+Lsv0S0k4imUfgehbdTdHMmxLnpTybFKFWRCYI03uuIJ0hjPS6dglR9Lu09R9WdpGZ2plYJ0lurzg3tudKMiC4lkAoep5q6ojGT81EIghQJQZDaPY+ixRrWoBx1tAIEqfNrBAAAptC/oVe89Q0vLj3Ks1bv5nElK7hj114ROxPvuf8hHvZMMS8rqxT+5pxsyo9/zJsqzvKmirNcfvxj4eOxkmVllTxs3Az+/Z87R9SlU7d8Hj9zVbUY1QrSpje25CZNmy+i9H2ALSXjL31fEdE6Cn/5CxLR+qqfab/QDyPxsg+xnliNuaLEqIoTBGm8S+z7d3/Q9CX2scZstyA1+/x2XGKvnRt1zOp/W9hBmkfhfzDJIyIfhZsHgdSAxkzuQyEIUiQEQWr3PIoWa1iDctTRChCkzq8RAACYItYb+vp9b1RLtqK5a7nzY/3YF2gfIeXa3fw77j1Y4Xnr9/DBM18Kf7OWMQdOX+A5a3dz78EKB26+LWL+fYH23KX7Ezxx7vooMarmiaeeEfEB9i8y/tI3nqLF2tNE9J3mb44a/A3i7BjtHv03EU0iAzGqYkeTJkqwazKRINX+jT5aIWnm77XSUv+YeIJU+1irj4v1/EbHTiZGz2X2eBpBmkdE1xLRxbHOBWALaMzkPhSCIEVCEKR2z6NosYY1KEcdrQBB6vwaAQCAKeK9oe9943Ne+cKrPHtNebVwG/ZMCXd6JD/ism41Dz8+kMdMD/LirYd5/6kvhL95uzH7T33Bi7ce5tHTg/zw4wOj5tgXaM+dHsnnYc+UxJSis9eU88oXXuV9J//No6bMT/cHWF0y/sJ3gYzl2s0U7m6t/duRKR4jsJc9ZGLHqB6nCFIjEajKyFh/r+8kr70M3azo1O7QtCJIzTy/0bHtkKT657EgSEFq0TdmepBq5h6NmZyLQhCkSAiC1O55FC3WsAblqKMVIEidXyMAADCFmTf0fSf/zavLQxESrnjFTh49bRHnDxnDf+7cnQPtfxcl8+75cxfu99Q4nhbcyBsOnuIjb38r/A3dSTny9re88cBJnhrcwP2GjeN7/twlag4DN/+O7+/yGOc/OYbHTF/EJSt3xhSjs1bv5jV7jkfIaQGCNI+Mv/Adp5ov7AEiuq5qLPUp3PV+r+Zvv6VwR0TgfPLIohhVMStIEXcGgjRtoDGTe1EIghQJQZDaPY+ixRrWoBx1tAIEqfNrBAAAprDyhn7g9Je89sUTPGfdi4ZybvzMVTxImcoP9xrEd973QJTsa9vuFu7UrQ/3Hfo0K1Pm83NLtvKy7S/x9r+dF/5Gn8qU/e0cLyur5GeXbGFlynzuO3Qsd3qkN7dpd0vUHN1x3wP8cK9BPGjUNMN7ihplzroXee3e1/jA6ejbHAgQpLEaNO2hmi/sVxk87hoiekPz958Rvti7gQ1kUYyqQJDKHQjStIHGTO5FIQhSJARBavc8ihZrWINy1NEKEKTOrxEAAJgi2Tf0Xa++z2v2HI8pS9WMnraI+w4dxw883Itv+d3vo4SgNrfd/Sd+sFs+5w8Zw4WT53HRnFU8Y1kpL9iwj1fueoU3V5zhF459IPxDofIc865j7/PmijO8ctcrvGDDPp6xrJSLZq/iv06ay/lDxnCnR3rzbXf9Me7rvfn233PHrr2437BxPGb6IlMyVCtF1+w5zrtefT/+B1j6BWkPMv7Ct5Vqdo/GojWF712pPuYfRPSjFI4V1B7LYlQFglTuQJCmBTRmcjcKQZAiIQhSu+dRtFjDGpSjjlaAIHV+jQAAwBS1fUM/+s53vOe1T3nL0X/yyhde5Xkb9scVe8XLt/MzM1fwyIlz+Inh4/nRfsO4w18e57v+0JHb3XJHXKGo341665338b0d/sIPdsvnLj36c9deg/jRvkO554AR3Huwwn2HjuX+wydwQeEkHjJ6Gg97pphHFM1mZcp8HlE0m4c9U8xDRk/jgsJJ3H/4BO47dCz3Hqxwj/4j+NG+Q7lrr0HcpUd/frBbPt9z/0N86533Ge76jJV2t9zBd/2hIz/wcC9+rN9T/MTw8Txy4hx+pmQlz1ixw5IQnbdhP6984VXecvSfvOe1T/noO9+Z+wBLvyAtIOMvfKurnj8rweN7UvgSe/Vxa1I1UCAWCFK5A0GaFtCYyd0oBEGKhCBI7Z5H0WINa1COOloBgtT5NQIAAFOk4g19/6kvePvL7/Hz+0/ykm2VlkTgrNW7ecLsNVw4aR4PHj2V+zw5lh/r9xR36taX/9CxK99+7/1802/Ni9RU5Kbf3sG333s//6FjV+7UrS8/1u8p7vPkWB48eir/dfJ8njB7jeXXrM+SbZX8/P6TvP3l92rV8EqAIC2k+F/4miZ4/EVE9JzuscNTM1QgEghSuQNBmnLQmMn9KARBioQgSO2eR9FiDWtQjjpaAYLU+TUCAABTpOsNfc9rn/LWo2/z6vIQL9x0qNYCcdbq3Txt0RZ+pmQlK1MX8oiJc3j4+Fk8bFwxPzn2WR48eioPLJzM/UcUcb/h47nv0Kc5f8ho7lUQ3iHaq0Dh/CGjue/Qp7nf8PHcf0QRDyyczINHT+Unxz7Lw8YV8/Dxs3jExDmsTF3I42eu4umLttgybn0WbjrEq8tDvPXo27zntU/t/QBLvyBVKP4XvkYmjnEZEe3SPBZNmyQEglTuQJCmFDRmkgOFIEiREASp3fMoWqxhDcpRRytAkDq/RgAAYApRb+hH3v6W9538N79w7EMurXyHNxw8zat2H+OlZS/x/I0HUiIiRWX+xgO8tOwlXl0e4g0HT3Np5Tv8wrEPef+pL/jI29+m9gPMeYK0gcnjoGmT5ECQyh0I0pSCxkxyoBAEKRKCILV7HkWLNaxBOepoBQhS59cIAABM4eQ39P2nLnD58Y95+8vv8Zaj/+QNB0/z2hdP8Iqdr/Di0qPCRer8jQd4celRXrHzFV774gneeOgMbzn6T97+8ntcfvxj3n/qgvA5dKAgtfL8aNokMRCkcgeCNGWgMZM8KARBioQgSO2eR9FiDWtQjjpawcnfpyFIAQDAAjK8oVe8+TXvP/UF7znxCe8OfcS7jn3Au159n3f+41+8/eX3uOzv7/K2l85xaeU7vPXo27zlyFu8+fCbvKniLG8+/CZvOfIWbz36NpdWvsPbXjrHZX9/l7e//B7v/Me/eNer7/OuYx/w7tBHvOfEJ7z/1Bdc8dY3wl+z6Q8wdwtSIjRtkhYIUrkDQZoy0JhJHhSCIEVCEKR2z6NosYY1KEcdrSDD92lL3y8hSAEAspJpb+iZFgkEKZo2SQoEqdyBIE0JaMwkFwpBkCIhCFK751G0WMMalKOOVsi079MQpAAAacm0N/RMiwSClAhNm6QEglTuQJDaDhozyYdCEKRICILU7nkULdawBuWooxUy7fs0BCkAQFoy7Q090yKJICVC0ybpgCCVOxCktoPGTPKhEAQpEoIgtXseRYs1rEE56miFTPs+DUEKAJCWTHtDz7RIJEiJ0LRJKiBI5Q4Eqa2gMZOcKARBioQgSO2eR9FiDWtQjjpaIdO+T0OQAgCkJdPe0DMtkglSIjRtkgYIUrkDQWoraMwkJwpBkCIhCFK751G0WMMalKOOVsi079MQpAAAaXHiGzoZf2FgIuLg5gpTxyitfKf6MWaeq7TyHcvj0z/mwe79kzpeSj/A5BOkaNokCRCkcgeC1DbQmEleFIIgRUIQpHbPo2ixhjUoRx2t4MTv0yn9fglBCgCQFSe+oZMLBalWjpodY1o+wOQTpETGTZtusuG4II2cXlb3rGiJh6QuoWCd8wRBWlvQmEluFIIgRUIQpHbPo2ixhjUoRx2t4MTv0yn9fglBCgCQFSe+oZMJsZko6RSkAwqnOFKOVp6TVpASRTdt+oAgC1zF6wsu++SbM2XCRR5if74+s41D8y77jCBIawsaM8mNQhCkSAiC1O55FC3WsAblqKMVnPh9OqXfLyFIAQCy4sQ3dDIpSLVikoj4we79q38XS5A2aNys+ucTZq+ptSB1shytPCe1ICVC0yZXc3rJ1TsuHF8kXOYh9ueL0AI+NOMXFQRBWhvQmEl+FEqRIJ1ZsIJXFm3lnUsOIi7IqomlXFKwnOcrq3nSU8W8ZMo64WNyY1YWlfLMghXCxRrWoBx1tIL6fbpztz6sTJkvfTo/2geCFAAgJ04WpPpoBaj2knZt8m6/L6Yg1cpRfZIRpFo5qh2bkyK5ICWKbtq02ubjgxTx2oKLh7y3s9NXomUeYn/e3dHh6yWD6DkKr3mv4FPNraAxk/wolKodpAOXc0kB4ppU1Wty/jwe0GEkj3lsuvgxuTUDxe88xBqUo45W8Hj9isfrZ48vkDnx+iFIAQDykeMNPOXxBXhE0WzhMs+sINXKT3XXpv5nekEa3FwRJUNru4NUHyfuIB1RNJs9vgA3adZiHskpSNG0yaXwbLrktXkX/Ve0zEPsz4m59N3FF1N7qmkkBKyhb8zUgdCYSUYUwg5SZAl2kNoVp+w8xBqUo45WwA5S59cIAABM4fH57/f4Atwtf4hwmacXkLF+r4pNdbeoGnVX6YDCKVGCVH2MfqcnaQSpVqISETdo3CyhIJ0we03188b6e5Hp2quAPb4AX3td40KSU5ASoWmTa3l94U8G/l95z49ECz3Evrz7wmOfLBv2I3X3aB4RNRd6krkPfWOmVYTGTLKiUKp2kBYs551LDvKrFa/zgR1HEIdnfUkZlxQs50VFa3jGpFm8dvEG4WNyY5ZNW89juk/nbnf2WUtEdye7hmoL1qAcdbSC9orM8hMf8+aKs1JnwMgibpGVy9c1bOSaGgEAgClyc3OvUrfKbznypnChl2pBqn8M1UKQTpi9JupnTrrUflPFafb4Apzr9X/1k5/+9A8kryAlQtMm1/LS7J+vfW93ty9Fiz2k9nnvhYe/fGHSz0qpRubdRER1RJ5fLgSNmTIHhVIsSPfvOMKrgusQhyc4YTWXFCzn4sKFXDh0ND83aabwMbkxk4cX84AOI/kef4cyIuqY7BqqLViDctTRClpBuqniLM9eWy51Hsl/kps0bc516tZ3TY0AAMA0ub7ASI8vwB0e6sH7T10QLvYogSB1yiX22sc4rVnT/pP/5j93frTq8vqWcyl9u48UEiNIidC0ya18b+/0y3f9355eX4gWfEgt5OjuHp/tmfrL7VTzXqMGmCdRY6afiRoYSAkKYQcpsgM7SO2KU3YeYg3KUUcrYAep82sEAACWyGnddr/HF+A/d370u9Kj/3S0IK0854wmTfrHqMcXfan95sNv8h86PfKdxxfgVjmtj1KkrPhlik8lJcYcByk9O1jRtMmd1Fk14kfPnpj7/f+e2/anCxeOB/nrM9uESz8kdr4+s42/CC3gc9v++MWJufSd7rL6PApfFo7do9aI15jpBmGjAqlCoRQL0rdC7yIuyK6lh7ikYDmvKynjVcF1vH/HEeFjcmPUeRR970qsQTnqaAUnNj1OZUZNXYB7kAIA5KZp06ZXZeX63lAvt+8/fALvOva+kDddMiFIK89F7tqkqkvr1d8ZCdLKc5GS1M4dpJXnInepaseSrux89V/cb9i46u6CWbm+Ez/60Y/uopov2O2I6JIUn0oKiRWkaNrkXhr/4GLKWzyYZhx87mcHQ3Mv++zYwjrnT8wlRpyVULDO+dDcyz478NzPDi4ZRM9pGjKpaUdEjcSdSq4EjZkyD4UgSJEQBKnd8yharGENylFHK0CQOr9GAACQDPWat8hZpwo2jy/Aw8bN4N3HPxL+RozETnnoQx40agpr69asRdZqir7UNR335FRIrCAlQtMmN3MDEQUo+txF3BM/ETUhYAU0ZspMFIIgRUIQpHbPo2ixhjUoRx2tAEHq/BoBAECytLy6Tp3uLbM9+3K9/u88vgC3CdzMfYc+zRsPnRL+hozUZOOhU9znyTHs87dXmzF91yLLs/fKa67pRtHSokWazh+FxAtSIjRtcjN1KCy0IUrdFbVeuKzeOmjMlJkoBEGKhCBI7Z5H0WINa1COOloBgtT5NQIAgGS5mIhyiCjvymuu6dYiK3eXdmfiPfc/xMOeKeZlZZXC35wzMcvKKnnYuBn8+z93jtgx2iIrZ/tVV/2qKxmLi2wK71BKBwo5Q5AShe/f97FmDGja5C7qENGNFN6R6CXxAhCJjpfC9bmRIEaTBY2ZMheFIEiREASp3fMoWqxhDcpRRytAkDq/RgAAUFsaUtUXtMuvvPLB5q1ySnO8/q+0Uq5d+99x78EKz1u/hw+e+VL4m7WMOXD6As9Zu5t7D1Y4cPNtEVI0x+v/qnnL7M2XX35lJ4otMBqk+bxRyDmClIioC6FpEwDAuaAxU+aiEAQpEoIgtXseRYs1rEE56mgFCFLn1wgAAOzg/xFRM9IIt980aDS8RavcHbk6WerxBfjhxwfymOlBXrz1MO8/9YXwN283Zv+pL3jx1sM8enqQH358IOvnOMfr/6pFq9wdv2nQaDjF39nVjIh+nM6TpQqFnCVILyI0bQIAOBM0ZspsFIIgRUIQpHbPo2ixhjUoRx2tAEHq/BoBAICd/JjCO1ny1Fx88cW31q1bL79xsxazWma33p/T2v+JXub9/k9duN9T43hacCNvOHiKj7z9rfA3dCflyNvf8sYDJ3lqcAP3GzaOf/+nLtFCtHXbj1tme/Y1adZ8Zt269XpffPHFv6X4YrQJib2MXCFnCVKisGTYqRkLmjYBAESDxkxAIQhSJARBavc8ihZrWINy1NEKEKTOrxEAAKSCHxBRI4psIFGdq676Vdfrrr9hXLMWWeuyc32v6WVfm8DN33Xq1of7Dn2alSnz+bklW3nZ9pd4+9/OC3+jT2XK/naOl5VV8rNLtrAyZT73HTqWOz3Sm9sEbv5OP0dZub4TzVpkrbvu+huejnNPUX3aVdXlByk/AxKjkPMEKRHRlYSmTQAA54DGTEAhCFIkBEFq9zyKFmtYg3LU0QoQpM6vEQAApJrLKbxb0VCWqqlbt15+4xtufK5FlueFHE/b83ohqM1td/+JH+yWz/lDxnDh5HlcNGcVz1hWygs27OOVu17hzRVn+IVjHwj/UKg8x7zr2Pu8ueIMr9z1Ci/YsI9nLCvlotmr+K+T5nL+kDHc6ZHefNtdf4z5Wj2+AGd72pxvkZW7q/ENNz5Xt2693vHm0SDtqub/8tSVOCkUcqYgJULTJgCAM0BjJkAEQYrohBAEqT3zKFqsYQ3KUUcrQJA6v0YAAJAuvkdE/0NEdSl830vtJYJRueyyy3531TXXPPKbBg0HN2rSbFrT5q2WtWiVW56V4z2e07rth/GEojZt293Ct955H9/b4S/8YLd87tKjP3ftNYgf7TuUew4Ywb0HK9x36FjuP3wCFxRO4iGjp/GwZ4p5RNFsVqbM5xFFs3nYM8U8ZPQ0LiicxP2HT+C+Q8dy78EK9+g/gh/tO5S79hrEXXr05we75fM99z/Et955H7dpd4up8VVdHv9hVo73eItWueXNWmQtbdSk2bTfNGg4+KqrftX14ssuuy3ePBmkbdX81q2a7++lppy1RiHnClIiNG0CAIgnSGjMBCBIEZ0QgiC1Zx5FizWsQTnqaAUIUufXCAAARPIjCl/SfB0RZZM1EZh3xRVXP1Sv3rUDGzZqMrZJ0+YlzVpkLW3eKresZXbrw1ke3xs5rdt8YFZSpiI5rdt8kOXxvdEyu/Xh5q1yy5q1yFrapGnzkoaNmoytV+/agVdccfVDVl+zQbKr5u9KcmjzO1sAACAASURBVNcuR4WcLUiNmjYVCB0RACCTyKXI9x80ZspcRpHx5+UCgiDNqECQ2juPosUa1qAcdbQCBKnzawQAAE7jf4ioDoV3x3ip9gIx7yc/+em9V131q66/rn/tE/V+02DwtddeP7RBw0Yjr7u+8aiGjZqMbdio6YTrGzed3Khps2lNmjYvvqFpi9lNm7WY1/TGlsGmzVrMu6Fpi9lNmjYvbtS02bTrGzed3LBR0wkNGzUZe931jUc1aNho5LXXXj+03m8aDP51/WufuOqqX3X9yU9+co8d4zaIt2peflU1T25GIWcLUqLopk3fEJo2AQBSDxozAS1FZPx5OZMgSDMqEKT2zqNosYY1KEcdrQBB6vwaAQCA07mIiH5MRD8nomuI6FoKS8IsImpDqRGRotKm6nXdUPU6r6l63T8i+RpxKOR8QUqEpk0AgPSjb8yUpwkaM2UeM8j483IqQZBmVCBI7Z1H0WINa1COOloBgtT5NQIAABn4IRH9lMIyqw6F5WJjImpORDkkXqS2qRpH86pxXVs1ziurxv3DJF6z21HIHYKUCE2bAADpw6gxU54maMyUeSwk48/L8RQ+J+olc1DIGfcFgtTeeRQt1rAG5aijFSBInV8jAADIJC6hsNj6Xwp/yfwFhb+M/pLCsvIqIrqawpew16Fwo6N6FH4zr1f133Wqfn911d9fWfX4y6uO97Oq4/+IiC5Oy6tyJwq5R5ASoWkTACA9BMkljZlaUffHs6kHx0sWdV+V7PGzqUdF+Bg9imo71gTjvC3JYzaoenyD2o4vAbvJ+PNyEIXPi7rJHBRyxn2BILV3Hmsp1hQiKowRrEH31DGtQJA6v0YAAACACBRylyC9iNC0CQCQWlzVmEkiQWr5ObKpx22ax6dakF4g48/L+yl8blyTzEEhZ9wXCFJ757GWYs1oTarBGnRPHdMKBKnzawQAAACIQCF3CVIiNG0CAKQO1zdmqu2OzFQSa2xZ1H1VMqIzjYL0ATL+rPyYas6Nq5I5MOSM+wJBau882iFICwsLq0MQpG6sY1qBIHV+jQAAAAARKOQ+QUqEpk0AgNTg+sZMRhJSuwtUv6tUJxk5m3pwK+r+uNFjDf6+QbaFHaDx5G3N89Tsdo03NqPfZVOPCoPnsrKL9ioi6krhz0Y1z1LsHWqlVMv70kLOuC8QpPbOox2CVItmfdY3c0ysQUfUMa1AkDq/RgAAAIAIFHKnICVC0yYAgL1I0ZgpniDVi0a94NSlgfaxMQSppXuJxvsbjbytqPrbuGOLJ0izqcd5o8dpxa8BrYnoS4p/ya4+A6jm/Ejq1guQM+4LBKm985hCQaquTaxB59cxrUCQOr9GAAAAgAgUcq8gJULTJgCAfQTJJY2Z4pFIkGr/VpWS2p2fBjs1YwpS3W7PhLtI4wlSzb1Uz1sYW9Ql9pqfaXeTRu1ONWAVWZOjJ6nm/GgT57hxKR64TJlZsIJXFm3lnUsOIi7IqomlXFKwnOcrq3nSU8W8ZMo64WNyY1YWlfLMghWOEKRYg8LrmFZUQdq5Wx9WpsyXPp0f7QNBCgAAAJhAIXcLUjRtAqlA0QRkBq5qzBSP+JfYxxaEuvuAmhKk2Zr7fur/zuzYVPSC1OTY4t6DVN/AKoEg1d571kw+I6I7KHyONIpz3LgUD1ymlAxcziUFiGtSVa/J+fN4QIeRPOax6eLH5NYMdMgOUqxB0XVMKx6vX/F4/ezxBTInXj8EKQAAAJAAhdwtSInQtAnYj+UmD8DVuL4xk5b4gjRSXsa7XN6MINUdK6GEtXiJvZmxxZK1ho9LIEhXkzVBykQ0nsLnyE/jHDcu2L3mvmAHqT3BDlI5gh2kzg92kAIAAADmUMj9gpQITZuAvUCQZhaub8ykxaIgPa+VktrHp1uQ6h9vcmxRglS729TK2Ci5e5AeIKKcOMdMiPb+h69WvM4HdhxBHJ71JWVcUrCcFxWt4RmTZvHaxRuEj8mNWTZtPY/pPp273dlnLRHdTcb/P6diZi3GEKTxomANprWOjkJ7D9LyEx/z5oqzUmfAyCJukZXL1zVs5JoaAQAAACJQSA5BSoSmTcA+tGuhPrlrHQBrSNGYSYtFQRpTOKZTkGp2j2rvJWpmbA0MHhc1DpM7SInCXewLKfwZqM0SMv6s/IjCu42/n+C4MdHKmf07jvCq4DrE4QlOWM0lBcu5uHAhFw4dzc9Nmil8TG7M5OHFPKDDSL7H36GMiDqSAwQp1mDK6ugotIJ0U8VZnr22XOo8kv8kN2nanOvUre+aGgEAAAAiUEgeQUoUbtr0DdW8jmVihwNciuVL9IBrCVJNrV3bmEmLRUEa1d0+TYI0xiXwEQ2ZzIxN3+m+QidbrVxir3IDRUpyNZ+S8eflHyksVpMCu9fcF+wgtSdWdpAWFhayHRQWFmIHqZg6OgrsIHV+jQAAAAARKCSXIL2IiCZR5GvpI3REwI1AkGYG0jRm0mJFkOr+nltR98drusdXX+qeLkFqdNl93LERRe0+Pa8di/q3mjFHNYAy4H+IqB1FC9LjZPx52ZeImpk4riFaOfNW6F3EBdm19BCXFCzndSVlvCq4jvfvOCJ8TG6MOo8J7l2pUGoEafVzYg2mpY6OQitIK8+x9Bk1dQHuQQoAAACYQCG5BCkRmjaB2qNUJUgQpLIiVWMmYDuXUnhXaH1NNpDx5+VfiahNsk8EOeO+QJDaO48QpO4OBKnzA0EKAAACyM7Ou9zj8V+f7fW3zvX5b/f4/A/m+Pz5uV7/iByvf3KuN7DQ4wtsyPX6y3N9gYMen/+oxxt4xeMNHPd4Aydzff63cn2B87newPser/9Tj9d/weMLcHW8/gser//TXG/g/Vxf4Hyuz/+Wxxs4WfX4Vzw+/9FcX+Cgx+vfXfU8C3K8/sker394bmt/7xyfv1Ouz397ttffunXrdg2zs/MuFz1nDkAh+QQpEZo2AXvIIwhSWZGqMRNIC9PJ+POyiMLnzaXJHBRyxn2BILV3Hs0I0kTRYubvCYI03XV0FBCkzq8RAAC4ge9n+XzX5bRue2dua/8TuV5/ca7Xv8Pj87+Z6/V/GyEzXZJcr/9bjzdw1uMLbM/1+Wfkev39cn3+27PbtGlAtWi64CIUklOQEoUvn0XTJlAb8giCVEaka8wE0oL+9i1qplP4vPnfZA4KOeO+QJDaO48QpO4OBKnzA0EKAAC1JCenbSOPz/+XXF9gYq7XX+rxBt5IJBtvan/rt/fedS93ub8j9364Gw/tlc/jCgZz8V8VXjy+iJ9/dga/MG8BH122nP++ajUfW7eeX9uwgU9v3sJvlW7jc9t38P/t2sUflpfzp3v38oUDB5grK6tz4cAB/nTvXv6wvJz/b9cuPrd9B79Vuo1Pbd7Mr23YwMfWree/r1rNR5ct511z5/Pzz87gRc9M4Bl/VfjpgYN4aK987v1wN+5yf0e+5857vr2p3S0JpW6u1/+6xxvYmusLTMxp7X/I4/FfL7o2NqOQvIKUCE2bQO2orwmQhyBJ1pgJpIXxZPx5OZPC584vkzko5Iz7AkFq7zyaFGv1ybh5WjxBavT3+kCQpreOjgCC1Pk1AgAAYdzo9/8svCu07SiPL1Dm8fk/iiUMb7/tTu7R5WEeO3AwL54wkfcsCPKZLVv5m8OHI2SmW/LN4cN8ZstW3rMgyIvHF/HYAYO4R5eH+fZb74gtTr3+D3O9/m0er1/J8ba9Izs7O6ldIw5BIbkFKZo2AQC0SNmYCaSF0WT8ebmAwufP1ckcFHLGfYEgtXceIUjdHQhS5weCFAAA4tDC5/tVrjfQy+PzL8/1+U8bScDf3/l7Luj+OM8ZNYbLZs3m4+ufj9rRKXu+2H+AQ+vX87aZs3jWqDE88LGefPftdxvvNPX5T3m8/mW53kCvFj7fr0TX2AIKyS1IidC0CQAQBo2ZQG1QKP7nZVKf/ZAz7gsEqb3zCEHq7kCQOj8QpAAAoCPHG2jr8foneLz+l/Vyr+Mf7+eRfZ7gJUUT+eiy5fzp3r3C5aST8+nevXx42XJePGEij+zzBD9w35+Mdpm+nOv1j8/xBtqKrn0CFJJfkBKhaRMAAI2ZQO1QKP7nZd1kDgo5475AkNo7j7UUa2YEKdag8+uYViBInV8jAACwHY+nbSDXF5jaum27d1VxF2h3y7dPdOvOi8cX8T/WrOWvKiqEC0cZ8tWhQ/yPNWt58fgifqJbdw5o7m/q9bV71+MNTHGoLFUoMwQpEZo2AZDJ6BszPUlozASsoVD8z8t6yRwUcsZ9gSC1dx4hSN0dCFLnB4IUAJCxNG/T5opcr3+Ez5/3jirobr751u+mDhvB/1izVrhIzKS8smYtTxr6FN/c/rff1ewsDZzN9fpHNG/T5grR50oVCmWOICVC0yYAMpUgRTZmyiE0ZgLWUCgFn5eQM+4LBKm982iHIFUUpToEQerGOqYVCFLn1wgAAGpF69at6+R6A0tVEdc20P67EX368b7gIuGiEKnkfcFFPKx3X27jb1+9szTXG1iSnR1IqqmDjSiUWYIUTZuAGRRNgPtBYyZgBwpBkCIhCFK759EOQRojeQRB6pY6phUIUufXCAAAkibX6x+hSreH7u/4benM2RnXUMktuXDgAJfOnM2d/tThm6p7lV7weP3DBZ4+CmWWICVC0yaQGO1aAO4mUWOmOqIGBlyHQhCkSAiC1O55rKVYU6oSNEgeQZC6pY5pBYLU+TUCAADLtPR663u8/kqPL8D33nnPN7vnLxAuABHz2T1/Ad97973/9fgC7G3T7qXsbL+IpkEKZZ4gJULTJhAfCFJ5SNSY6XuCxgXch0IQpEgIgtTuebRJrOXFCdage+qYFiBInV8jAACwRLbX39rrC3zu8QV46rARaLbk0nxVUcGThj6lXnL/SW6bNrlpPpUUykxBSoSmTSA22rVQn+ReBzKDxkzAThSCIEVCEKR2zyMEqbsDQer8QJACAKSmdevADR5f4FOPL8Bls+cIl3xI7bNt5qxwEyef/6OcnLaN0ng6KZS5gpQITZuAMZbvYQYcSZDQmAnYh0IQpEgIgtTuebRJrNWPE6xB99QxLUCQOr9GAABgmlxf4IjHF+D1058TLvYQ+7J26vSqnaT+Q2k8nRTKbEGKpk3ACAhS94PGTMBuFIIgRUIQpHbPo2ixhjUoRx2tAEHq/BoBAIApctu0a+fxBbjznx8QLvQQ+9PxD/d/6/EF+MZWub9N0ymlUGYLUiI0bQLRQJC6GzRmAqlAIQhSJARBavc8ihZrWINy1NEKEKTOrxEAAJgi1+vv5/EFeOKQYcJlHmJ/xg9+kj2+ADe+4cbJRPTDNJxSCkGQEqFpE4gEgtTdoDETSAUKQZAiIQhSu+dRtFjDGpSjjlaAIHV+jQAAwBS5Pn+ZxxfgRzp2Fi7zEPvz4J8eYI8vwC1zPEeIKCsNp5RCEKQqRk2bLhE6IiAKCFL3gsZMIFUoBEGKhCBI7Z5H0WINa1COOloBgtT5NQIAAFPkegNBjy/AN7e/TbjMQ+zPze1vY48vwM1bZG2i8AfY/6T4lFIIglSLvmnTArHDAYJQqhIkCFK3ESQ0ZgKpQSEIUiQEQWr3PIoWa1iDctTRChCkzq8RAACYwuP1K7leP3t8AZ4zaoxwoYfYlzmjxqhNmrQfYNek+JRSCIJUy0VENI7QtAmEySMIUjeBxkwglSiUIkE6s2AFryzayjuXHERckFUTS7mkYDnPV1bzpKeKecmUdcLH5MasLCrlmQUrhIs1rEE56mgFVZB27taHlSnzpU/nR/tAkAIA5MTj9Sseb4DbBtqzxxfgzcUzhYs9pPbZXDyTPb6qunrT+gGmEASpnsuIaCtFNm1qJ3REQBR5BEHqFtCYCaQahVK1g3Tgci4pQFyTqnpNzp/HAzqM5DGPTRc/JrdmoPidh1iDctTRCuHv0+ENRxmTyA049YUWAAAA7EL9F69BPXpXv+EN7tmLP9u7T7jkQ6zns737eHDPXtW1VOsKQSocNG0CROHzXw1wNv0JjZlAalEIO0iRJdhBalecsvMQa1COOloBO0idXyMAADCF+oY+Z/RYfn3DRr7nrnvY4wvwzTffxnPHPA1R6pJ8tncfzx3zNN98c/ieo/fcdQ+/vmEjzxk9FoLUOeQQmjYB4AYup7AURWMmkEoUSvE9SF+teJ0P7DiCODzrS8q4pGA5LypawzMmzeK1izcIH5Mbs2zaeh7TfTp3u7PPWiK6O9k1VFuwBuWooxW09yAtP/Exb644K3UGjCziFlm5fF3DRq6pEQAAmEIrSFXRpjwxsHoHoipKz+3YIVwCItE5t2NHhBj1+AKsPDGwWmxDkDqOBwlNmwBwOkFCYyaQehRKsSDdv+MIrwquQxye4ITVXFKwnIsLF3Lh0NH83KSZwsfkxkweXswDOozke/wdyoioY7JrqLZgDcpRRytoBemmirM8e2251Hkk/0lu0rQ516lb3zU1AgAAU+gFqVa89ezycMS9Rgb37MVbikuwq1RwPtu7j7cUl/DjD3WNqE/PLg9HiWwIUseBpk0AOBt9Y6YHCI2ZQGpQCDtIkR3YQWpXnLLzEGtQjjpaATtInV8jAAAwRSxBqhWlyhMDuX37Ww1lKXaWpifnduzgLcUlEfcX9fgC3L79raw8MTBmHSBIHQmaNgHgTIwaM/kJjZlAalAoxYL0rdC7iAuya+khLilYzutKynhVcB3v33FE+JjcGHUeRd+7EmtQjjpaQStIK8+x9Bk1dQHuQQoAkJNEglS7a3Fz8Uwe1CNS0Hl8Ab73rnt5yrDhvGdBEMLURiG6Z0GQpwwbzvfedW/UnA/q0Ys3F89MuJsXgtSxoGkTAM4DjZlAOlEIghQJQZDaPY+ixRrWoBx1tAIEqfNrBAAApjArSI1kqdHOUvW+pYN79uK5Y57ml1auwiX5JubzpZWreO6Yp3lwz14R9xPV7xQ1I0UhSF0DmjZlDoomwJmgMRNINwpBkCIhCFK751G0WMMalKOOVoAgdX6NAADAFMkIUn1e37CR54wawz27PGwoTNVdplpp+tLKVcLFpIior12VoUa7Q1Uh2rPLwzxn1Bh+fcPGpJ8PgtTxoGlTZqBdC8CZBAmNmUB6UQiCFAlBkNo9j6LFGtagHHW0AgSp82sEAACmsEOQ6hO+PHxhtTQ1EoDa3aaPP9S1Wp5uKS5x9a5TdTfoluKSagn6+ENdDXeF6hsszRk1hvcsWGjrbQogSB0PmjZlBhCkzgaNmYAIFIIgRUIQpHbPo2ixhjUoRx2tAEHq/BoBAIApUiFIjfL6ho28uXhmtTTt/KcOcYWhfvfp4w91jRCpavYsCFbvyjSKlTHGO86eBcGI51XH8/hDXWPuAjVK5z91qJahm4tn1mp3KASpNKBpk/xo10J9wjpwEmjMBEShEAQpEoIgtXseRYs1rEE56mgFCFLn1wgAAEyRLkEab8dl5YqV1fJUeWJg3Ev1nR710njliYHVErRyxUphO2IhSF2DUdOmK4SOCNiJdi3kVQU4AzRmAqJQCIIUCUGQ2j2PosUa1qAcdbQCBKnzawQAAKYQLUjN5NyOHVy5YiVXrlhZfem+mkE9enHPLg/HjBW5Ge84g3r0injePQsWVo/JzkviIUgzGn3TpkOEpk2yAEHqTNCYCYhEIQhSJARBavc8ihZrWINy1NEKEKTOrxEAAJjCDYIUgSDNINC0SU4gSJ1JkNCYCYhDIQhSJARBavc8ihZrWINy1NEKEKTOrxEAAJgCglTuQJC6DqOmTb2FjgjYAQSp80BjJiAahSBIkRAEqd3zKFqsYQ3KUUcrQJA6v0YAAGAKCFK5A0HqStC0ST4gSJ0FGjMBJ6AQBCkSgiC1ex5FizWsQTnqaAUIUufXCAAATAFBKncgSF0LmjbJhVKVIEGQOgE0ZgJOQCEIUiQEQWr3PIoWa1iDctTRChCkzq8RAACYAoJU7kCQuho0bZKPPIIgFQ0aMwGnoBAEKRKCILV7HkWLNaxBOepoBQhS59cIAABMAUEqdyBIXQ+aNslFHkGQiiZIaMwEnIFCEKRICILU7nkULdawBuWooxUgSJ1fIwAAMAUEqdyBIHU9aNokF/U1AekHjZmAk1AIghQJQZDaPY+ixRrWoBx1tAIEqfNrBAAApoAglTsQpFJwGRE9T2jaBEBtQGMm4DQUgiBFQhCkds+jaLGGNShHHa0AQer8GgEAgCkgSOUOBKk0/ILQtAmA2oDGTMBpKARBioQgSO2eR9FiDWtQjjpaAYLU+TUCAABTQJDKHQhSqUDTJgCSA42ZgBNRCIIUCUGQ2j2PosUa1qAcdbQCBKnzawSAFHDLlg05O5sT5Lxtz5eVtZazstbG/H2rVnfEG0stnjffztdhBQhSuQNBKh2dCE2bALBKkNCYCTgPhSBIkRAEqd3zKFqsYQ3KUUcrQJA6v0YASEE6BSlnZR3l7GyujSDl7GzmVq3usPS82dmT7Ra9VoAglTsQpNLxPULTJgCsgMZMwKkoBEGKhCBI7Z5H0WINa1COOloBgtT5NQJASqolZnb25JQd26Qgjfpddvb5ZEQnBCkCQQosgqZN7kPRBKQPfWOmRYTGTMA5KARBioQgSO2eR9FiDWtQjjpaAYLU+TUCQEpiCdKqS9RrdnEaSM6onZ5ZWfkGx427EzSeII14Du2xawRoTVq2bBjzd1WvzXC3qua4dgFBKncgSKUFTZvchXYtgPShb8zkJTRmAs5BIQhSJARBavc8ihZrWINy1NEKEKTOrxEAUmIkSKPkaI1MPFr9NzW7Ow0lqG2CVDe+mGOr2i0aS5DGvbVAlVy1CwhSuQNBKjVo2uQeIEjTDxozAaejUIoE6cyCFbyyaCvvXHIQcUFWTSzlkoLlPF9ZzZOeKuYlU9YJH5Mbs7KolGcWrBAu1rAG5aijFdTv05279WFlynzp0/nRPhCkADgBQ0Gqyk+N0NT+LEI2qjs3s7LW6nea1vYSe6PjVh9TFbHaXaHRu0irL7HX/Ez7OlOyixSCVO5AkEoPmja5A+1aqE9YB+kgSLEbMzUWNSgANCiUqh2kA5dzSQHimlTVa3L+PB7QYSSPeWy6+DG5NQPF7zzEGpSjjlbweP2Kx+tnjy+QOfH6IUgBEE3UDs3ETZzCf2ei23wqBGnE7/S7WOMIUsNjQpAiEKQgGjRtcgfa+uRVBaSORI2ZsNMaOAGFsIMUWYIdpHbFKTsPsQblqKMVsIPU+TUCQEqiBGnirvKJ/m5y1LFrJ0gjx2d8CX3iHaTxXhcEKQJBCiIxatqUJXREQA8EafpAYybgFhRK8T1IX614nQ/sOII4POtLyrikYDkvKlrDMybN4rWLNwgfkxuzbNp6HtN9One7s89aIro72TVUW7AG5aijFbT3IC0/8TFvrjgrdQaMLOIWWbl8XcNGrqkRAFIST5CaPoZOPkYduzaCVCMxdbtbo0Vt/Evsz3P0fVQhSBEIUhALfdOmc4SmTU4CgjR9oDETcAsKpViQ7t9xhFcF1yEOT3DCai4pWM7FhQu5cOhofm7STOFjcmMmDy/mAR1G8j3+DmVE1DHZNVRbsAblqKMVtIJ0U8VZnr22XOo8kv8kN2nanOvUre+aGgEgJTHuQRopIbVislWrO7SNkqofYyQq41web/S4OGM7H/M5tDtKa54330CQRsjQiGNBkCIQpMAYNG1yLhCk6QGNmYCbUAg7SJEd2EFqV5yy8xBrUI46WgE7SJ1fIwCkJIYgNb6M3UwXe+3f6O/1maCLfcxENouK/XfRgrSmi33N68Ql9ggEKbCCvmnTPLHDAVVAkKaHINXMMxozAaejUIoF6VuhdxEXZNfSQ1xSsJzXlZTxquA63r/jiPAxuTHqPIq+dyXWoBx1tIJWkFaeY+kzauoC3IMUACdgJEirfp6vk4jRTZL00lEjR6v/JoGITCRIE/59y5YNNZfyVx8/YmxVY9ePpVoEx9nhmgwQpHIHgjTjQNMmZ6JUJUgQpKkCjZmA21AIghQJQZDaPY+ixRrWoBx1tAIEqfNrBAAApoAglTsQpBkJmjY5lzyCIE0FaMwE3IhCEKRICILU7nkULdawBuWooxUgSJ1fIwAAMAUEqdyBIM1Y0LTJmeQRBGkqQGMm4EYUgiBFQhCkds+jaLGGNShHHa0AQer8GgEAgCkgSOUOBGlGg6ZNzqO+JsAe0JgJuBWFIEiREASp3fMoWqxhDcpRRytAkDq/RgAAYAoIUrkDQZrxoGkTkJ0goTETcCcKQZAiIQhSu+dRtFjDGpSjjlaAIHV+jQAAwBQQpHIHgjTjQdMmIDO3EBozAfeiEAQpEoIgtXseRYs1rEE56mgFCFLn1wgAAEwBQSp3IEgBoWkTkJPvU3jHKBozAbeiEAQpEoIgtXseRYs1rEE56mgFCFLn1wgAAEwBQSp3IEhBFb8golcITZuAPKAxE3A7CkGQIiEIUrvnUbRYwxqUo45WgCB1fo0AAMAUEKRyB4IUaGhGaNoE5EDfmGkYoTETcB8KQZAiIQhSu+dRtFjDGpSjjlaAIHV+jQAAwBQQpHIHghToQNMmIANBimzMpN09isZMwC0oBEGKhCBI7Z5H0WINa1COOloBgtT5NQIAAFNAkModCFKg43tE9DShaRNwL2jMBGThrwRBioQgSO2eR9FiDWtQjjpaAYLU+TUCAABTQJDKHQhSYMAlhKZNwJ2gMROQicFk/Hm5hiBIMyoQpPbOo2ixhjUoRx2tAEHq/BoBAIApIEjlDgQpiAGaNgE3gsZMQCbyyfjzspQgSDMqEKT2zqNosYY1KEcdrQBB6vwaAQCAKSBI5Q4EKYgDmjYBN4HGTEA2upLx5+WLBEGaUYEgtXceRYs1rEE56mgFCFLn1wgAAEwBQSp3IEhBAtC0CbiFIKExE5CL35Hx5+WbFD6vr0vmoJAz7gsEqb3zKFqsYQ3KUUcrQJA6v0YAAGAKhD0BZAAAIABJREFUCFK5I0CQjiLjL3zz0/T8wBpGTZseFToiAKJBYyYgI78m489LJqI/ElGjZA4KOeO+QJDaO4+ixRrWoBx1tAIEqfNrBAAApoAglTsCBGkRGX/Zm5am5wfW0Tdt+pLQtAk4BzRmAjLzLhl/Zo4joqbJHBByxn2BILV3HkWLNaxBOepoBQhS59cIAABMAUEqdwQI0hlk/GVPSdPzg+RA0ybgVNCYCcjMYjL+zHyXiOZQ+LNTm7xEB4SccV8gSO2dR9FiDWtQjjpaAYLU+TUCAABTQJDKHQGCdCEZf9kbUPX89VL8/CB59E2bDhIuXQZiQWMmIDux7kMaK0qiA0LOuC8QpPbOo2ixhjUoRx2tAEHq/BoBAIApIEjljgBBupuMv9Q9UvX8v0rx84PagaZNwEkECY2ZgPwcpxQI0lWTSnnX0kOIC7Jq4jYuKVjO80et5snDi3np1PXCx+TGrJpU6gixhjUoRx2toH6f7vxYXx49dYH06fxYXwhSAICcQJDKHQGC9ALFbjiRR0S/TPHzg9ohtGlTaDVdemIODXpj8c+3vr7g0o9PLat7+sRcYsRZObWs7unXF1z68RuLf771xFwq4Nkp2WmMxkwgU/g9mRek0yj8Of7TWAdT5QzirswsWMHje5bwgA4jecxj04WPx80RLdawBuWooxXU79OZFghSAIB0QJDKnTQL0gfI+Avdv6hGbMT8Ugccg5CmTW8s/tmQ1+Z//z/v7Xzg8y+PL+ZvzpQxn69EHJpvzpTxheOL+b2dHT5/bd5F/z256CcDbTwd0JgJZBrPkTlBGqQEn+UzB63IKx64THm6+3PThnYeG0Tck6535K+9x9+hrOsd+WtFj8XNeeDmrv0TrZNUgjUoRx2tkOP153m8fqVlVu60pje2DGZKfl3/WtfUCAAATAFBKndSJEjzKLpxxLMU+wvdMqoRGz+24flB6klr06aTi3+5+oO9+R+Jln5I8vlwX/6HJ5f8YpFNpwQaM4FM5Gki+oRqKUg11KfIe/Yizk9HCr//dXTAWGRIfRJLfRI/BzKkPrmH+iR+vkSkPgEAgAxAkMqdFAlShaw1lehGNR+gl9rw/CA9pKVp06mlV697/8XHL4gWfEjt86/ybp+/seTKFbU8JdCYCWQqv6DwOf44EY0jovkUFqLaWNmtU5/Ef2lGrOVuCsvRux0wFhlSn8RSn8TPgQypT+6hPomfLxGpTwAAIAMQpHLHAYL0Zar58PTZ8NwgvaS0adPJJb986v0Xe78vWuwh9uVf5T0/eCP480G1OC2ChMZMIHNpQPZ9Gf1p1d8hSKZG9G2dsAblqKMVMrXmbqoRAADEBoJU7ggWpF8SUQeq+UJ3rQ3PDdLL9yhFTZtCq+nS1+Zf8h/RQg+xPyfmXfQNK3RREqcFGjMBQHQ1EbUl7NYBAAAAAAAgfUCQyh2BgvQzIupJkV/ocP9Rd5KSpk3H51385Hu7On0lWuYh9ufd7R3+s0Gh8RZPCTRmAiCSnxDRlURUj7BbBwAAAAAAgNQCQSp3BAnSvRS58yuPiFra8LxAHLY3bTq19MqdF44vES7zEPvzRSjIh4t/cZCI6lo4JdCYCQAAAAAAAACAGHJ8/nyPL8DjBg0RLvMQ+zN24GD2+ALcqMmNU8k+QZpHYUk6jWoaR8wnovFE1J2iLwUMENFlNjwvEIutTZteX3DZJ9+cKRMu8xD78/WZbRyad9lnRHSTydMBjZkAAAAAAAAAAIgjt027dh5fgO+9617hMg+xP3fdfjd7fAGuU+83fcj++5bVp8SNJNpS+DJBIAf6pk3FyR7o9LK6Z0SLPCR1CS2s8y4RtSNzl8YHNefUEUJjJgAAAAAAAAAA6SbXGzjs8QV41ZSpwoUeYl+WT5zMHl+AW+X6jlFqGjvUp/hytCkR/cDG5wPiMWra9FAyBzoxl4RLPCR1OTGXmMLvA80TnAr6xkx3ERozgcwmh4j+QkQjKHy1hjZ5YoYEAAAAAABABpDVOtDc4wuwxxfg7bPnChd7SO1TOnM2e3wBzvX6/3PVVb/qSjXC4RobT52fUnTjiHpEdBUR/T8bnwc4C1uaNkGQyh2NIPXHOQ3QmAmAGh4jovMU/x7fiqjBAQAAAAAAkBE0z/E8qkrS4fn9+NyOHcIlH2I9b5eV8bDefVmt5W8aNBpOkbs6/0fgaQbkodZNmyBI5Y5GkHrjnAbxGjNlERozgcxhIsUXo6YF6fN5LfPW5WUpCJKpeT6vZV6qFqoZsAblqCMAAIAMp2GjZoNzvW2/8PgC7Gt703cThwzjj8r3CJd+SOJ8WF7O4wc/yd62N7HHF+Acr//z+tde9yRFylHLu/wAiEMzIvo/Mt+0qaX2PyBI5Y5GkObFOB8SNWbCvYtBpnAPmZOjTOHmiPUpfAWHIevyspT17bMYcVeed8AYZMm6vCwlpSs2AViDctQRAAAA+PHll1/ZqWWW54C6AzHQ7maePPQpPo8dpY7M+R07eNLQp9gfaF+9a7RFlmfvz3/+ywcoUja0I6IfiTu1gKTcS+abNu0hzZd6CFK5Y0KQBjXnDRozgUzlYiI6SeYFaZAS3E9clTMv9nyQX500BnFBXuzRmde3z+IXe3QWPhY358WenR0h1rAG5aijFXK8/jyP169kWnK8/jzRcw8AAKnmCiLKu/Kaa7q1yPLsVaWbxxfgIT1786658/nzffuEi8FMzuf79vGOOXN5SM/erK1Pi1a55br7jWpzuZCzCciO2aZN9at+l6f+AIJU7iQQpGjMBECYjmRejloSpMcmjeGvXw8hLsixyWN5ffssPjZ5rPCxuDnqPM7xNUu40zqVYA3KUUcreLx+RfudLGPi9Sui5x4AANLBL4koQER5V1xx9UPNW+Vu078h9ujyMC+eMJHfLisTLgwzIW+XlfGiZyZw985/ifpwat4qd9vlV1zRmYzFqJ/C94sEIFVcQkTrKX7TpiDp7p8HQSp34ghSNGYCoIalZCxCvyKidRR+79SmP/1/9u4+Tq6qzvf9757rPXI8em50GEXEsRgEmSRC0p2EBoJWLpzX6KhHrjMggkoEJPIww4ORQRlMERQBQ4AmCCFPnY7AQGLIiB0gBDo8SewgIJBABwKxBxCYiTCO6EyOss4fVatr1a61d+1dtav22qs+79fr99JUV+9atdf67a76sncVAal3pQOhRy65UD177wb12pbNmY8pj6X34/d7Dxxo1CftRA/6MY9J6ID0hJPPVKWFy7yvE075WwJSAF1nDyl/xmBRRIrvec+ffn7iR6cs75lx6LPBgO4LnztWXX/RxeqJ1WsyDxJ9qidWr1HXzbtIff7//5u6ULRnxqHPTpx88LL3vOdPjxV7MFoUkUki8vZOLxx0pQkS/qVNBeP2TfoXCEj9roiAlC9mAqpeE3tAeqmE/20vCgGpV6UDoQfnnac2rhpQ2+/ZkPmY8liuBGv0oB/zmIQOSC9auEw98pLyvi66crmacegRakrP9Nyc5QsAafnvUg7airr2fN/7jp84+eBlU6cdsmV638zfBcO7k77wRXXVBReqDTcs4bNLY9ZLGzaoDTcsUVddcKE66QtfrAtEp/cd/ubUaYds+YvJBy/d833vO16i3zhNFD5vFJ0X9qVNVxm3va7vTEDqd4UEpHwxE1D1QbGHo7+T6L/xRSEg9aoISNPdj1kHa/SgH/OYRLcGpJM+OmVAcjJHAJC2/yblL84oBuv9++zz1Y9MnHz1lN7pw70zDns5GO795f/85FvfmHOaGrjkUrXlppvVS10emr501wY1ctPNasX3LlNzTz1N/eX//ORbwX3We8hhL03pnT78kUmTr9p7731Ose13Sx1YmScgK8EvbVoi5VDUfPNfECEg9b1CAtIBYx3wxUzodkWxB6TbpNoXR4jIh6V83DSr4bfYE87kpwhI092PWQdr9KAf85gEAan7cwQA7fJ2EfmQiEyTkKDuHe9612cKf/7hv580ecrKqdP6Rqb3Hf6bYADYd9jH1P/61P9Sc754oiqdda5aXJqvfrzoB+UA9a4N6g8jI5kHmc3UH0ZG1It33qW23HiT+qdF16rFpfmqdNY56tQvnqg+81efUX2Hfaz+7NBDDv/NlN5Dfjbpo1MGCn/+4b9/x7ve9ZmwfWupaZX54FJ6uMD2pU3Bmi1CQOp7WQJSvpgJqBX2BU2bpNobeyXdKOFM/oqANN39mHWwRg/6MY9JEJC6P0cA0Al7SDmcO1gqX+oUVu997/u/dMCBE783+aCpN0/pmXFfz/S+rdP7Zr4R9c14Hy8e9dZnP/VZ9cVjjlOnn3iSOv+0M9QlX/+GuvbbJTV46eXqtv5F6p6ly9WWG29Sj91yq3rqR2vVM+vWqR23/0T9cv0d6qW7NqhXN25Uu4aH1b/df7/63UMP1YSZv3voIfVv99+vdg0Pq1c3blQv3bVB/XL9Heq5229Xz6xbp5760Vr12C23qi033qQ2LlmmbutfpFZ+7zJ17bdL6rvnzlXnn3aGOv3Ek9QXjzlOffZTn1Uf//iRdWeC1l4mP/ONnul9W6f0TN806aCpN334gImXvPe97/9S1H6z1BEicpCI/Fll/wOu+bCIbJHwgPQqEQJS3ysQkPLFTEC9U8V+jByS6t/7xAhn8lcEpOnux6yDNXrQj3lMgoDU/TkCgCy8U0Q+IOXPLDUvnYysP93rAyd+aN/95h4wcdIVkw6aeuPBPdM39kzr+0XvjMNenN53+JtRoaMrNb3v8Dd7Zxz2Ys+0vl8c3DN946SDpt54wMRJV3xo3/3m/uleHzgx7r6w1CGV/fmByv4FXDJBRD4uIvNEZFjqL6cPOzuKgNTzCgSkfDETUO/rYj9G3irV3kis2XBGRNSkA/bPPBxJWnq/3Xz1lTW3TzpgfyUimY8vSSAUFpDq5yIi6u9O/HLNz26++srxnzUzf5896kj14K03pzIHwQrOSbvH40qwZuvBF+6/N3Q/tbr/G82NuS70Y7YyJ81sI4/zmAQBqftzBAAueJuIvEfKB82DpPaMoUT17j33PO4DHyyctu9++19wwIGTF0ycfPCyj07pvW1Kz/RNPdP6Hp06/dCneqcf+kzv9MN29M44bGfvjMNe7D3ksFenHXL4v047ZOYb0w85/M3pfTP/I3BG539MP+TwN6cdMvONaYcc/q+9hxz2au+Mw17snXHYzt7ph+3onX7oM1OnH/pUz7S+R6f0Th/+6MG9aydOPnjZAQdOXrDvfvtf8IEPFk579557Htfsc7LUzMp+KlT229ua3fFABxSk/JmSb0jjUDRYBKSelxGQHiN8MRNgM0/sx8cBKffGpGY22q0BaXDsvgakwef5dyd+uemAVP9uuwJSEVHf/+Z5HRuPK8Fa0oBURNQL99/btv5opa/TWiN5nMckCEjdnyMAcNV/E5H3icifS/nFf4+IHC7phYx5qMMrz3tSZT+8T/hyJeTXBBEpSbKgtEhA6ncZAem9xrzzxUxAVUmiA9KPNLPRNALSB2+9WYmI+uxRR6rPHnVkzRlj3//medYzBPXvSEToYwZ9OjAKhjfm9oOP0SicM8O4YEAaNb5mnm8zY20UCDUKSPX/Bs/mM38ed3zm89TPPe48xgnhzDNb48xBGuNxJViLCkjDAm69dvV+MINv/XzjrDezx/QcNDqDNLjvG60R2zaCYwsG4+batT2Wi/OYRNyAtPiJo0NfGxc/cXTmwWfxE0ergds3E5ACgCPeISLvFpG9pRwcThSRqVK+zDzyM04dqiMq451aGf+fV57PuyvPD/DVBCl/AdNOaRyQnh0nIG20nRd/vt5639U3XFZ32/WXfWv8/8cJ+F78+fpE96esAem5gTnji5mAqpJEB6T7NbPRNAPSYJnhiy4z/LHdPyxssYV6ZihkVtRZiMFx6TDJDEgbjS/p8212rK0GpDqE0pfZ63HrsSTZl7bwK848xg1Izcd48NabG247jfG4Eqw1E5DqsNPWJ3HXW/B3g9uwhZu234maE9s2wsZmfhyErZfMbbo4j0n4EJAe/9WzlYgQkAJAzuwh5Usy/0RE3i/lL4naX8pnY06R8pmZvSIyXURmSDmsPFREDpPymZtHiMjHpDbU/Fjl9sMr9zu08nszKtvprWx3SuVx9q887vsr43iX8IVJgGm2iDwuIS8CRWQgzYD07K8eX3P75p8M1N3245XVs1lcCEiP/kRxPMjN07bj1lOLRYnIL405GBS+mAkwlSTk+CgtvOFMOyANBoz67EX977Cz+czfD27PPONMP2ZYgDTpgP0jAzr9O/psOR24NLrEPmp8jZ5vs2NtNSDV4wyGZsHb444vzuXTwXlLEpDqOQ47sza47VbH40qw1sol9uYZpI3CVXM+bT3W6AzS4Lox50DPmW1OzG2Yz0s/B1v/mCF/2GO7No9JJL3Efv0jL9aFogO3b65bF+sfebEuXNVBpvnz/Q6cXHObiKj9Dpxc85jnzFtYs+3LFq8ODW4bhbUEpAAAAMkVpfylTMEXfS8kCUjNM0XDAkERqQkFbbe5Uu0cmyvP+/zP18z3b6T8H5+KleKLmYAcBKTm2V16fMHQLuoyen1/HdQEv2DIfEzzkmxbRQWxwTPebr76ytCA1Da+pM+32bG2GpAGw9JJB+w//gU6zezLqEAyaj/p0o9n/v84Aalt282Mx9VgLWlAaq4X86zbYNAZ9ftxeiy4rsN+x6xGAWnYNoJzbzteBMfm2jwm0WpAqv8dLDPkDIaY+me224O/a4aqZp0zbyEBKQAAQIdNkfKb/vEXXz+7Op2AVAeCuiYfuJ/1trAzQoP31Y8Vdv/gmanBIFLfvvknA9bt2sYW5/fiPH7YtpupuOOx1diWn6j/vkfNi3C+mAmoVxJ74DEgjgakZphhBh62M9z080krIA07q9A2bn2bfvxG40v6fJsdaxoBqXk5vUj5jLykAakeXzD8irufzH0cFXTZLrGPutw76XhcDdaSXGIfts/MIDHOfOYlII0am2vzmEQaZ5AGSyr72XYGqe2s0+CZojogDTujdL8DJ1tDVC6xBwAA6IyClN/8v7FybmuX2J/91eNbDkiD9zN/brt/MJzUtbB0TsMx67CyUUAa9ntxHr8dAWnUeMLqa1/+a/N3npPyx5UUhS9mAkwlsffZgOQsINVhSPDyWZH6y3CDAUqjy8KTjNsMdMzHazS+NALhVipJQBoMyx689ebYl9g3Cr8a7ae4c2D73TjbbnU8rgRraQekcX43To8Fw82oj77Q+7zVS+xtnwkctW5cmsck0gxIzcvlReoDUjPAvGzxaiVSG5rqcFWHn/o+YaUv0ycgBQAAyMaEtRemE5CawWDUJfbBwNM8MzIYQC4snVN3f/3vYEAY3Ib+tw4tbY9jG2+j34v7+GldYh/neYTVmzseVH/3WVEi8nsROV/4YibApiSeBKRRZ7gFgx9dti9pivMlM3HCueDjNxpf0ufb7FjTCEhtz8/2eY5xxmeGyZ896shY8xg2B2Glg7Y42251PK4Ea2kHpHHnM60vaTIfPzgntm0k+ZImAtLoS+zNMzpF7AGp+bmkaQSkOhAlIAUAAMhIpz+DNBh4rr7hMiVSG7SaFXb/sNr8k4HQMQdviwpIw34v7uM3CkiDl8yHnREa53lEVeVb7D8nIn8rfDETYFMSez8PSM4C0mDQ8v1vnjd+Jpp55pn5OZJhAUkwbLF9y3WjkMUM1mxBUHB8zTzfZsaaVkAa/CKfsC+8aTS+4EcSxJ3HuAFp8HcabbvV8bgSrLUjII273swea/QlTWEhqfnYtjmxbUPPixhzFbWGCUhrA1J9Wbz+TFDbpfO2gNT87NKkl9gHi4AUAAAgI74HpFHPIyogDfu9rALSZubDCEi/JCJnS/nF87S2LCQgv0riUEDazrJ9q3XUt1h3WzUKSKlk+zHrYM3FHsxTuTKPSbQakEad5RkVkJq367J9SVPwPuZ9gwGpOS4CUgAAgA7IOiCNusT+7K8eH/sS+2aCxWYC0riPn/Yl9i0GpP+fiHxeRI4QvpgJCCpJlwSku0frv4VcV/BMs24sAtJ092PWwZqrPZiXcmUek0jjM0jNEPOceQvHzwTVZ3SGBaTBzy0NXmJvC0CDjx08a7XR2aYEpAAAAClKEpCGVdRZk40CUvM+wTIDyTj3N0PL4O9EBaTm7yb9vbDHt207w4C0WKm3pb1+AA+UpIsC0t2j9ZdiR317djcVAWm6+zHrYM3lHsxDuTKPSSQNSNMq8xJ7HaTqoLNRyNlKEZACAACkyIWA1BY66vAv7P7Bb5I/+hPFxMGiefZqkoA0zuPbtu1AQAqgXkm6LCClogMhAtJ09mPWwRo96Mc8JpFVQBo8e9Qs/XmmBKQAAACOixOQUvktAlKgoZIQkFKjBKRp78esgzV60I95TCLLgPSRl1Td35Hgt9oTkAIAADiMgNTvIiAFGioJASk1SkCa9n7MOlijB/2YxySyDkg7XQSkAAAAKSIg9bsISIGGStKmgPS2WT1q01ePV08uuJjKQd136glq7awedfuXPqeWn3aSuvNb52Y+pjzWplOPV7c5EKzRg37MYxIEpO7PEQAAgLMISP0uAlKgoZK06wzSYo9aO4vKTVXma9XMg9TFUw9Q1/dNyn5Mea1i9sEaPejHPCZBQOr+HAEAADiLgNTvIiAFGioJZ5BSCziDNK1y5cxDetCPeUxCB6QnnHSmKi1c5n2dcPKZBKQAAABpISD1uwhIgYZK0ubPIH1ty2b17L0bKMfroXnnqbWzetQdXz9d/fDKBWr4psHMx5TH2nDBXHX9oZPVORP3XSMin262h1pFD/oxj0nM6JtZmtE3U8049Ijuqb6ZBKQAAABpICD1uwhIgYZK0uaAdPs9G9TGVQOU4zV07mlq7awedcucE9UV3zpf/XDhgszHlMdacfrJ6uKpB6jj9937ThE5rtkeahU96Mc8JsEZpO7PEQAAgLMISP0uAlKgoZJwBil1L2eQplWunHlID/oxj0mYn0E6/PTr6vbNL3hd51x4uTq4Z7r68P4fyc0cAQAAOIuA1O8iIAUaKkmbA9Ldo1upHNRTV3xHrZ3Vox6cd57auGpAbb9nQ+ZjymPp/Zj1Z1fSg37MYxJmQPrjzS+oxWuGva6vnPH36i8mHaT2+WAhN2f5AgAAOIuA1O8iIAUaKgkBKTVKQJr2fsw6WKMH/ZjHJDiD1P05AgAAcBYBqd9FQAo0VBICUmqUgDTt/Zh1sEYP+jGPSZgB6SMvKe/roiuX8xmkAAAAaSEg9bsISIGGSkJASo0SkKa9H7MO1uhBP+YxCQJS9+cIAADAWQSkfhcBKdBQSQhIqVEC0rT3Y9bBGj3oxzwmQUDq/hwBAAA4i4DU7yIgBRoqCQEpNUpAmvZ+zDpYowf9mMckCEjdnyMAAABnEZD6XQSkQEMlISClRglI096PWQdr9KAf85gEAan7cwQAAOCsHTd+8IWsQzyqfbV1YJ+XhYAUiFISAlJqlIA07f2YdbBGD/oxj0kQkLo/RwAAAM4aXf72N/7383dmHuRR6dfu5+9QW5e+/TdCQApEKQkBKTVKQJr2fsw6WKMH/ZjHJAhI3Z8jAAAAZ+1Y9f4Nv9u2MvMwj0q/3ty6XD286E82CwEpEKUkBKTUKAFp2vsx62CNHvRjHpMgIHV/jgAAAJz1zPK3nffK3V/4j6zDPCr9+tWGY3evmivXSPlFc1/GSw1wVUkISKlRAtK092PWwRo96Mc8JkFA6v4cAQAAOEstlv/nmaX/5Q9Zh3lU+vX0EnnrbW+TWVJ+0Twz46UGuKokBKTUKAFp2vsx62CNHvRjHpMgIHV/jgAAAJw2uuJd5746POfXWQd6VHr1q3tOeePGb75Dnz1aFJGDMl1kgLtKQkBKjRKQpr0fsw7W6EE/5jEJAlL35wgAAMB5P1/8njWv3HvS77MO9qjW65V7Tvz9PQvevV6q4ejHRWSfLNcX4LCSEJBSowSkae/HFoO1kojMCyl6MD/z2FEEpO7PEQAAQB78X/dfvefGVzed9mbWAR/VQjh676m/2XTln94l1XBUFwC7i8QekC4XAtKuKgLSdPdji8GarSd10YP5mceOIiB1f44AAADyYp9b/uEd/U8v+b//8NIdf/27320bULufvyPz0I8Kr93P36He3LpcvXTH5958eom8FbisvigihwtnjwJRLhd7CHOdEJB2VRGQprsf0whI582bN15CQJrHeewoAlL35wgAACBPDtzjbVIc/IYs+uk17/7p1iVv/81TK/Z5+ekloii3auvAPi9vXfL23zx0zbt/umquXGN8IZOuj4nIR7JbSkAuLBJ7QHqlEJB2VRGQprsf0whITUZvFuJskx50Yh47ioDU/TkCAADIm4kicoTUX6ZN5admishfCIBGVog9IL1Uyr30oWY2SjiTvyIgTXc/tjEgLVaKHnR/HjuKgNT9OQIAAMijfaT85T4EpfkqPV9cVu8RNWXK/qq3VzWol1N7vJ6eNaqnZ03oz6dO/WTUWFp43DPSfB4x3Sv2gHSulHvpg81slHAmf0VAmu5+JCDNdxGQul8EpAAAAJ21j4h8VMpnJPZJ9gEgVV99Up6fjwrBqJc6GZCqnp4tqrdXtRKQqt5epaZO/WSix+3tvSLtoDem34k9ID1Gyv21dzMbJZzJXxGQprsfCUjzXQSk7hcBKQAAAICuNh5i9vZe0bZtxwxI637W2/tyM0FnRgHp58Uejr4u1RBmr2Y2TDiTvyIgTXc/EpDmuwhI3S8CUgAAAABdLSwgrVyiXj2L0xJy1p3p2dNzhmW7kWeCRgWkNY9hbrsagFZrypT9Q39WeW7Ws1WN7ca0l4jMFpGSUf1iD0eViKyXagjz7oSPJSKEM3ksAtJ092ODYK0U0X/jFRKQRlWJHuzoPDqFgNT9OQIAAACA1NgC0rpwtBombhm/T/XsTmsImlo9qNtOAAAgAElEQVRAGhhf6NgqZ4uGBaSRHy1QCVdjOEREfi/xwhVd50g1IH17zMep8aNiT+m2WT1q01ePV08uuJjKQd136glq7awedfuXPqeWn3aSuvNb52Y+pjzWplOPV7c5EpDSg22fR6cQkLo/RwAAAACQGmtAqsNPI9A0b6sJG/WZmz09a4JnmrZ6ib1tu+Pb1EGseVZo/Vmk45fYG7eZzzPpWaS3SLJw9FmphqOHxXyMOj8q9pTWFnvU2llUbqoyX6tmHqQunnqAur5vUvZjymsV4wWk8+bNU2mYN2+e/QxSerDd8+gUAlL35wgAAAAAUlN3hmbjL3Eq3y/Gt823IyCt+VnwLNaIgNS6zeQB6YgkC0iVlD+btCgiH4n5GHU4ey1/xRmk6VSSM0jbEJCOPyY92JF5dIoOSE846UxVWrjM+zrh5DMJSAEAAAB0r7qAtPG3yje63xV1224tIK0dn/0S+sZnkEY9r/gB6a2SPCBdJuU3mxNiPkYd8/MPX9uyWT177wbK8Xpo3nlq7awedcfXT1c/vHKBGr5pMPMx5bE2XDBXXX/oZHXOxH3XiMinJcOAlB5s+zw6ZUbfzNKMvplqxqFHdE/1zSQgBQAAANCdogLS2NsIhI91224lIDVCzMDZrfVBbfQl9i+r+s9RTRqQNvMZpM+KyLSY27cyw5nt92xQG1cNUI7X0LmnqbWzetQtc05UV3zrfPXDhQsyH1Mea8XpJ6uLpx6gjt937ztF5Djp7GeQDoglIKUH2zaPTuEMUvfnCAAAAABSE/IZpLUhpBlMTp36SfOLksZ/xxZURlweb/u9iLG9HPoY5hml1cc9wxKQ1oShNdtK9k32e4nIPCmHJ2atEnvI8p9SfrO5R4LHqMHZa/krziBNp5KcQdqo0gpI6cG2zaNTzM8gHX76dXX75he8rnMuvFwd3DNdfXj/j+RmjgAAAAAgNSEBqf0y9jjfYm/eJ/hZnw2+xT60ar8sKvx+9QFp9Vvsq8+zlUvstYlS/fIls/5N7EHLcSLyoYSPMc4MZ3aPbqVyUE9d8R21dlaPenDeeWrjqgG1/Z4NmY8pj6X3Y8zPriyE9GVUQGq7f7AK9GBH59EJZkD6480vqMVrhr2ur5zx9+ovJh2k9vlgITdn+QIAAABAamwBaeX2MwIhYv2XJAVDRyMcHb9PgyCyUUDa8P5TpuxvXMo/vv2asVXGHhzLeBAccYZriP8hIh+T+iBlm9gD0m9IC5fZE87krwhI092PBKT5rrwHpJxBCgAAAACA3X+V8uX2BaPWiT0g/Z6UA9WmEM7krwhI092PBKT5rrwHpI+8pLyvi65czmeQAgAAAABScbXYA9IrpIXPISWcyV8RkKa7H1sM1uIEpPSg+/PYUQSk7s8RAAAAAMBNC8QekF4j5Tec72lmo4Qz+SsC0nT3IwFpvouA1P0iIAUAAAAApOVSsQeki6X8hnOvZjZKOJO/IiBNdz+mEZCWSqXxEgLSPM5jRxGQuj9HAAAAAAA3zRd7QLpcym84925mo4Qz+SsC0nT3YxoBaUgVhYA0L/PYUQSk7s8RAAAAAMBNJbGHMANSfsO5TzMbJZzJXxGQprsfWwzWSpUasFRRCEjzMo8dRUDq/hwBAAAAANxUkuiA9M+a2SjhTP6KgDTd/ZhSsFaMKHowP/PYEQSk7s8RAAAAAMBNJYkOSD/UzEYJZ/JXBKTp7kcC0nwXAan7RUAKAAAAAEhLSaID0kIzGyWcyV8RkKa7H1MK1goRRQ/mZx47goDU/TkCAAAAALipJASk1CgBadr7MetgjR70Yx6TICB1f44AAAAAAG4qCQEpNUpAmvZ+zDpYowf9mMckCEjdnyMAAAAAgJtKQkBKjRKQpr0fsw7W6EE/5jEJAlL35wgAAAAA4KaSEJBSowSkae/HrIM1etCPeUyCgNT9OQIAAAAAuKkkBKTUKAFp2vsx62CNHvRjHpMgIHV/jgAAAAAAbioJASk1SkCa9n7MOlijB/2YxyQISN2fIwAAAACAm0pCQEqNEpCmvR+zDtboQT/mMQkCUvfnCAAAAADgppIQkFKjBKRp78esgzV60I95TIKA1P05AgAAAAC4qSQEpNQoAWna+zHrYI0e9GMekyAgdX+OAAAAAABuKkmbAtLbZvWoTV89Xj254GIqB3XfqSeotbN61O1f+pxaftpJ6s5vnZv5mPJYm049Xt3mQLBGD/oxj0kQkLo/RwAAAAAAN5WkXWeQFnvU2llUbqoyX6tmHqQunnqAur5vUvZjymsVsw/W6EE/5jEJAlL35wgAAAAA4KaScAYptYAzSNMqV848pAf9mMckdEB6wklnqtLCZd7XCSefSUAKAAAAAEhFSdr8GaSvbdmsnr13A+V4PTTvPLV2Vo+64+unqx9euUAN3zSY+ZjyWBsumKuuP3SyOmfivmtE5NPN9lCr6EE/5jGJGX0zSzP6ZqoZhx7RPdU3k4AUAAAAANCykrQ5IN1+zwa1cdUA5XgNnXuaWjurR90y50R1xbfOVz9cuCDzMeWxVpx+srp46gHq+H33vlNEjmu2h1pFD/oxj0lwBqn7cwQAAAAAcFNJOIOUupczSNMqV848pAf9mMckzM8gHX76dXX75he8rnMuvFwd3DNdfXj/j+RmjgAAAAAAbipJmwPS3aNbqRzUU1d8R62d1aMenHee2rhqQG2/Z0PmY8pj6f2Y9WdX0oN+zGMSZkD6480vqMVrhr2ur5zx9+ovJh2k9vlgITdn+QIAAAAA3FQSAlJqlIA07f2YdbBGD/oxj0lwBqn7cwQAAAAAcFNJCEipUQLStPdj1sEaPejHPCZhBqSPvKS8r4uuXM5nkAIAAAAAUlESAlJqlIA07f2YdbBGD/oxj0kQkLo/RwAAAAAAN5WEgJQaJSBNez9mHazRg37MYxIEpO7PEQAAAADATSUhIKVGCUjT3o9ZB2v0oB/zmAQBqftzBAAAAABwU0kISKlRAtK092PWwRo96Mc8JkFA6v4cAQAAAADcVBICUmqUgDTt/Zh1sEYP+jGPSRCQuj9HAAAAAAA3lYSAlBolIE17P2YdrNGDfsxjEgSk7s8RAAAAAMBNJSEgpUYJSNPej1kHa/SgH/OYBAGp+3MEAAAAAHBTSQhIqVEC0rT3Y9bBGj3oxzwmQUDq/hwBAAAAANxUEgJSapSANO39mHWwRg/6MY9JEJC6P0cAAAAAADeVhICUGiUgTXs/Zh2s0YN+zGMSBKTuzxEAAAAAwE0lISClRglI096PWQdr9KAf85gEAan7cwQAAAAAcFNJCEipUQLStPdj1sEaPejHPCZBQOr+HAEAAAAA3PRtISClRglI096PWQdr9KAf85gEAan7cwQAAAAAcNM3xB6QrhYC0q4qAtJ092PWwRo96Mc8JkFA6v4cAQAAAADcdIbYA9L1QkDaVUVAmu5+zDpYowf9mMckCEjdnyMAAAAAgJtmiz0gvU8ISLuqCEjT3Y9ZB2v0oB/zmAQBqftzBAAAAABw01+KPSDdKeU3nB9uZqOEM/krAtJ092PWwRo96Mc8JkFA6v4cAQAAAADc9GdiD0iViHxORD7SzEZ1OHPfnBPU1iu+Q+Wg7p9zglo7q0f95EufUwOnn6w2XPD1zMeUx7qvsh+zDtboQT/mMQkCUvfnCAAAAADgrl+JPSC9REQmNbNBHc5Q+arbZvWo5YdNVhdPPUBd3zcp8/HkubIO1uhBP+YxCQJS9+cIAAAAAOCuQbEHpL8SkRtEpBSoYqMN3lacUvxRsad0w6GTr/p+74EDVH7qnIkfWnP8vnvfec7ED63Jeix5rlM+/IGzJcPQhh70Yx6T0AHpCaf8rZp/5XLv64RT/paAFAAAAACQmrDPIQ2rUoJtF6T8xpXKTx0nImdX/jfrsfhQBclWQbLfBz5UQRynA9JuKwJSAAAAAEBatgkBKVWuT0s5HP20A2PxoQqSrYJkvw98qII4blrfzOKMvpmlKT3Tr5r00SkD3VJ/Vvjz3JzlCwAAAABw22ckfkB6lZTfiE6Isd0JlftSVLdWnD5pJ3rQj3lMoiDZB8pZVEEAAAAAAGjRNRIvIB0Q3owCgKsKkn1YmUUVBAAAAACAFHxXRN4QAlIAyKtuPWs4T2f5AgAAAAAc9idSDj+/JiKXiMgyKQeiZrX6eW9FEZlnqWKT2wOAoKJ073GmKP49d9vzmZfpiAAAAAAAXttP2ns5Y0la//InAIhSku49zpTEv+cedkUDAAAAAABt834ROVwISAHkU0m69zhTEv+eOwEpAAAAACAz7xKR94nIhyS9z3srif2N7lUtbhcAtJJ073GmJP4997CAtCB8HjYAAAAAIIdKwpc/AWivknTvcaYk/j33sIC0KPn+bFUAAAAAQJcqif2N7hrJ75t3AG65SvwLCeMqiV/H2AlCQAoAAAAA8Mxssb/R/XcR+Z7k7807ALecKCKvi/04s0jyGRImMVv8OsZeKfbn84oQkAIAAAAAcmqKhJ8NpGtYRFaIyDyq6+rjRsX5rMQJxv2zHjuVXa2Q8nFDNahTxP+AtF3H2LMkWW/q+56V8HF0DUt40K1E5EEhIAUAAAAA5Njj0vgNPEUpEXlMymHJFCmbIOUzBG+T6PCEooJlnnFYEL914hj7upTP7ixUHvOzUg5dX+jAYysRuVQISAEAAAAAOXa0ZB+WUPmrF4RQlGq+/kGqgdo+4rdOH2Nf6PDjPScin5byXPals8sAAAAAAOi8OZJ9YEJRVHeU/uxRXe8U//l6jP2tVD8qoSgik9PZXQAAAAAAZON4KZ8JlPUbboqi/KxXpPbM0aJ01xmHvh1jHxeR46R2PvdKZ1cBAAAAAJCNPaX8BvcUERkQv97IUxSVTf1WRO6U2s+oNGtP6R4+HGP1fJpnjeqaJiJvS2E/AQAAAACQqb2k9g3vp0XkbKMGqK6rOyX+l8z8tnLfBx0YN5VdXSrl44X+XMqw8v2zR23SPMY+KPF78/FKJdm+rn+ojM0WiuqaKd3xUQkAAAAAgC7xTimfCVSkqEBdKuUQ1BbADDgwPiof1SfddeZoUNrH2OOkHJaGBaNpPU5YTRGRPWI8bwAAAAAAcuedIlIQwlKqttYIASmVvGaKyIHCZ1Sa0jzGni2dDUj1fHLWKAAAAACga7xNRCYYVaC6rg6U8pliA9I4IJ1Zue9kB8ZNZVd7Sfl4wedSNtbKMXaylC9/jxOQTqlUku3r2rMyNkJRAAAAAADQ1UoSHZAWshkW0NWKEh2QTslqYAAAAAAAAL4pCQEp4JqiEJACAAAAAAB0RNhnHS4SAlIgK1PE3pcPCgEpAAAAAABAqgpSH8L8VsrfpF0UAlIgKy9LfW/+QAhIAQAAAAAAUrdIqgHMS1I+q7QoBKRAlv5Kqn35nyJyk4jMEQJSAAAAAACA1H1NqkHM3VI+e/QzUg5i9sluWEBXMy+z1//hQgekk7MbFgAAAAAAgH9mSzWIGZFyEHOCiMwUkT2yGxbQ1Q6Ual++KrUB6Z7ZDQsAAAAAAGTqzIXDirJXC7t1ttR+CcwpIvJJEZnQ+owBaFJBqn35L1Luy+NFZK+4G8j6mJTiMQoAAAAAAGhnLhxW21/5AxWoFAPSO6V8htoBrc8WgBYUpNqXr0i5L6cl2YBLx0sCUgAAAAAAUuLSG36Xqg0B6YGtzxaAFhSkPiCdkWQDLh0vCUgBAAAAAEiJS2/4XapK+FCsVFKzhYAUcE1B6gPSviQbcOl42eIxCgAAAAAAaFm94f/2d99Svb1K9fYqdf+jtT/Tt59+1lt5DR+Ok2oQc68QkAIuKEiHA9Ko41yrxzoCUgAAAAAAUpLlGVFHHlUOB445To3fdvpZ1UAhq3GlED4cLbVf0lQUAlIgawXJ4AxS23EuGJ5mcIwCAAAAAABalgHp6qE/jgcEi5b+se7fBKQAUlSQDAJS87i2eqh8XLv/0T+0fKwjIAUAAAAAICVZf6aeecZo2JlWngSkE1OaMgDNKUhGn0Gqj3NHHqVq/t3KsY6AFAAAAACAlGQdkG5/pXomVdRn9XkQkE5OacoANKcgGX5Jk/l5o2kc6whIAQAAAABIiQsB6aKl1UtQv/3d7L6YiYAU8FpBMgxIzc8cTeNL6AhIAQAAAABIiQsB6THHqZrL7LMeDwEp4KWCZBiQbn+l+oVNaXwJHQEpAAAAAAApyTogNc8edeks0hbDh09INYgZEQJSwAUFyTggTeOzRwlIAQAAAABIWdYBqXm5qXkJatafQ9pi+FCUahDzuBCQAi4oCAEpAAAAAAAIyjIgNb+sRN/myjfZE5AC3ikIASkAAAAAAAjKKiBdPVS9tH7R0j82vN2DgPSgdGYMQJMKQkAKAAAAAACCsgpIo84UtZ1Z6kFAOiWdGQPQpIIQkAIAAAAAgKCsP4PU1SIgBbyzl1T78teSQUDq0DEKAAAAAABoLr3hd6kISAHvTJBqX/5WCEgBAAAAAICIW2/4XaoWw4c+qQYx24SAFHABASkAAAAAAKjn0ht+l6rF8GGKVIOY54SAFHCBLSA9LMkGXDpeEpACAAAAAJCSMxcOK8peQkAK+MQWkM5MsoGsj0kpHqMAAAAAAIBFkQqtpGwBaU8T2wGQnpYDUkPRsQIAAAAAACkoUm0NSKc1sR0A6SEgBQAAAAAAkYoUASngMQJSAAAAAACADjlQqkHMmBCQAi5IMyAFAAAAAABAhIJUg5hXhIAUcAEBKQAAAAAAQIcUhIAUcA0BKQAAAAAAQIcUpD4gnZHheAAQkAIAAAAAAHRMQeoD0r4MxwOAgBQAAAAAAKBjCkJACriGgBQAAAAAAKBDCkJACriGgBQAAAAAAKBD9pJqEPNrISAFXEBACgAAAAAA0CG2IIaAFMgWASkAAAAAAECHEJAC7iEgBQAAAAAA6BBbEHNYlgMCIHtItS93CwEpAAAAAABA23CmGuAmZVRR6EsAAAAAAIC2ICAF3ERACgAAAAAA0AEEpICbggHpxzIdDQAAAAAAgKf4rEPATcGAtJjlYAAAAAAAAHzGpbyAewhIAQAAAAAAOqQkIvNFZEAISAFXlCo1IASkAAAAAAAAbXeR1J6xRlFU9lWSajhaFAAAAAAAALRNSbIPgyiKqq2SEJACAAAAAAB0REmyD4MoiqqtkogUjAIAAAAAAECbFaT2jDWqO+osS2U9JqpaBQEAAAAAAEBHFCT7MIjqfNnOXsx6TFS1CgIAAAAAAICOKEj2YRDV+SIgdbsKAgAAAAAAAKBtbAHphEoBAAAAAAAAgNeiziAFAAAAAAAAAK8RkAIAAAAAAADoWgSkAAAAAAAAALoWASkAAAAAAACArkVACgAAAAAAAKBrEZACAAAAAAAA6FoEpAAAAAAAAAC8VBSRAREpRNwnSUBaEJEVIT8DAAAAAAAAAOdsknLouULsQWmcgLRQ+X1V2R4AAAAAAAAA5EJRasPPYFAaFZAWpBqMmj8DAAAAAAAAgNzYJPUhqA5KbQHpcSJyh+X2TR0dNQAAAAAAAACkoCj2IDRpFTs7bAAAAAAAAABIxyZpLRzd1OkBAwAAAAAAAEBaisLZowAAAAAAAAC62Cbh7FEAAAAAAAAAXaoonD0KAAAAAAAAoIttEs4eBQAAAAAAANClisLZowAAAAAAAAC62Cbh7FEAAAAAAAAAXaoonD0KAAAAAAAAoIttEs4eBQAAAAAAANClisLZowAAAAAAAAC62Cbh7FEAAAAAAAAAXaoonD0KAAAAoEmzRWSYoiiKoigq5/W61IajrzswJorqppotAAAAOVWSeN/+SlEURVEURVEUFVYlAQAAyKmSZP9iiqIoiqIoiqKofNeAlD/WoiAAAAA5U5LsX0xRFEVRFEVRFOVHLRAAAICcKUn2L6IoiqIoiqIoivKjrhIAAICcKUj5Upg3pfqi5iIR+aaInE1ROarfC2uY8q/mSnVd/28pr2vWNuVDccymKL9qo9QGpBNEZA8BAADImf+U6ouauVJ+oXOklMNTispD7RbWMOVffVqq6/r3Un0jmvW4KKrV4phNUX7VGqn29MrKbQUBAADIGfOyGP0GnDcqVJ6KNUz5WASklK/FMZui/CoCUgAA4AXzjcqnK/UnUr48hqLyUKxhysf6kFTX9W+lvK7/yoFxUVSrxTGbovyq66Ta09dJOSD9MwEAAMgZ841KsVL/JcPxAEmxhuGjCVIbkBZFZGaWAwJSwjEb8MtVUu3pRVLu6X2yHBAAAEAzbG9UgDxhDcNHBKTwFcdswC8EpAAAwAu8UUHesYbho72kuq5/LQSk8AfHbMAv10u1pxcKASkAAMihPaT6gma38EYF+cMahq8KUl3brwgBKfzAMRvwz4BU+/pSISAFAAA5ZLuEs5jheICkWMPwVUEISOEfjtmAfwaEgBQAAOQcb1SQd6xh+Kog9QHpYRmOB0gDx2zAPwNSH5DuneF4AAAAErO9UflYlgMCEmINw1cFqQ9I+zIcD5AGjtmAfwakPiDdK8PxAAAAJMa3JCPvWMPwVUEISOEfjtmAfwaEgBQAAOQcb1SQd6xh+KogBKTwD8dswD8DQkAKAABybi+pvqD5tfBGBfnDGoavDpTq2h4TAlL4gWM24J9/lGpfzxcCUgAAkEMF4VuSkW8FYQ3DT1OkurafEwJS+KEgHLMB36yTal//gxCQAgCAHCoIb1SQbwVhDcNPBKTwUUE4ZgO+ISAFAAC5V5D6NyqHZTgeIKmCsIbhJ1tAOiPLAQEpKAjHbMA3toD0vVkOCAAAIKmC8CUgyLeCsIbhJ1tAOi3LAQEpKAjHbMA3toB0zywHBAAAkFRBeKOCfCsIaxh+IiCFjwrCMRvwDQEpAADIvYLwRgX5VhDWMPzUJ9W1vU0ISOGHgnDMBnxzp1T7+jwhIAUAADl0oFRf0IwJb1SQP6xh+Koo1bX9uBCQwg8cswH/bJJqX58tBKQAACCH+JZk5B1rGL4qCgEp/MMxG/DPJiEgBQAAOccbFeQdaxi+Kkp9QNqT4XiANHDMBvyzSeoD0vdkOB4AAIDEbG9UZmQ5ICAh1jB8VZT6gHRKhuMB0sAxG/DPJqkPSCdkOB4AAIDE+JZk5B1rGL4qCgEp/MMxG/DPJiEgBQAAOccbFeQdaxi++oRU1/aIEJDCDxyzAf9slmpfny4EpAAAIIf6pPqCZpvwRgX5wxqGr46W6tp+UAhI4QeO2YB/HpdqX58iBKQAACCHisK3JCPfisIahp8ISOGjonDMBnxDQAoAAHKvKLxRQb4VhTUMP9kC0oOyHBCQgqJwzAZ8YwtI/0eWAwIAAEiqKPVvVHoyHA+QVFFYw/CTLSCdnOWAgBQUhWM24BtbQPrOLAcEAACQVFH4lmTkW1FYw/ATASl8VBSO2YBvCEgBAEDuFYU3Ksi3orCG4afjpLq27xUCUvihKByzAd88I9W+/rIQkAIAAIcdLSLzpP4D0z8h1Rc0I1L/RqUgIqW2jw5ojDUMXx0tIidabp8t1bV9p9QHpIXKfQAXccwG/HOVlHs0aKdU+/o4qQ9IZ4f8HgAAQCZ2isjrInKWcVvYtyQXRGRF5fbZnRsiEGmnsIbhnwki8oaIvCC1QelssQekBamu7WLHRgkkt1M4ZgM+KUm5R1dIbeC5U+wB6Wwp/23b1KkBAgAAxDFbqi9e9Bvx4BuV40Tkn4zbdmYwTiDMbGENw08lqV/bs6U2ID1bRNYat23q/DCBRGYLx2zAJ/o/6Ol+1UHpTuO2k0TkmsBtxU4PFAAAoJGdUn2xokTk1ZD/r2t2FoMEIuwU1jD8E3zTqUTkX4z/v0vq13Yxi4ECCe0UjtmAT0pS37f/bvz/3wZ+timLQQIAADQyW+pf1ITVzkxGCESbLaxh+Kkk8df2pkxGCCQ3WzhmAz6x/Qe9qCpmMkoAAIAY4r6oOTqrAQINsIbhoyRvOovZDBFoCsdswC8l4T/mAQAAD5SEFzTIt5KwhuGnkrC24Z+SsK4Bn8T9D3rFjMYHAAAQS5wXNcWsBgfEwBqGr1jb8BHrGvBPSaJ7+v7MRgYAAJBAScJf0GzOblhAbCVhDcNPJQlf2w9nNyygJSXhmA34pNF/+Dgqu6EBAADEF/WippjdsIDYWMPw1QThLDv4h2M24J+rhI/MAAAAHigJL2iQbyVhDcNPA1K/ttdlOSAgBSXhmA34pCD2gHRKhmMCAABIzHY2RzHLAQEJsYbhq4LUv+EsZDgeIA0cswH/DEhtTw9kORgAAIBmlYSzOJBvJWENw08DwhtO+KckHLMBnxSE/5gHAAA8YJ7NUcx2KEBTWMPwVUF4wwn/cMwG/DMg/Mc8AADybevQMVOeXn/Mx7u5Ju43YcWE//FfH8t6HC7U1qFj2v6ZSay59Is13LjSXtus487U3u99xx17v/cdd2Q9Dl8rzb6gJ+IXx+zk1YnXJ7CjtxvX986d/nkRUd87d/rnsx5LlkWfAgBybdv6Ywe2rT9WdXNtvvVoNXBpMfNxOFIDrLn8FWs4VqW6trexjjtSd6/4lLp7xacyH4fHlVpfbKMnYhfH7KYqtbWKZLbR27GKnj5WbaNPAQB5tm39sUfrP2o77/0C1aVlvLA5mjVH+VTtWtusYyrP1Y6+oCeodlQnX5/Ajt6mGhV9CgDwxrb1x7yxbf2xavfWU5Ua/RrVZbV766lq2/pj1dahY/5NRP7fSr2TNUflvdq9tlnHVB6rnX1BT1BpVhavT2BHb1NhRZ8CALyyrXLpzK4tX1Fq++lUl9WuLV9R29YfqzYs/9SdUv7CiKKIHMiao/Je7V7brGMqj9XOvqAnqDQri9cnsKO3qbCiTwEAXtGXzuzY+AWltp9JdVnt2Fi+NObyb/T9g1Rf2OzJmqPyXu1e26xjKo/Vzr6gJ6g0K4vXJ7Cjt6mwok8BAN4Zv3Rm2xlKbf87qktq97Yz1Lb1x6onbv+b30r1Rc1M1hyV9+rU2mYdU+dObaoAACAASURBVHmqTvQFPUGlUVm+PoEdvU0Fiz4FAHipeunMyUo9exbVJbVry8mZXRbDmqPaWZ1a26xjKk/Vib6gJ6g0KsvXJ7Cjt6lg0acAAC9VL505Xqlnz6G6pHZsPD6zy2JYc1Q7q1Nrm3VM5ak60Rf0BJVGZfn6BHb0NhUs+hQA4K3xS2eeOVup575OeV67nzk788tiWHNUO6rTa5t1TOWhOtkX9ATVSrnw+gR29DZFnwIAusL4pTOPnKrUc3Mpz2vXI6dmflkMa45qR3V6bbOOqTxUJ/uCnqBaKRden8CO3qboUwBAV6heOnOCUjvOozyvHRtPyPyyGNYc1Y7q9NpmHVN5qE72BT1BtVIuvD6BHb1N0acAgK5RvXRmrlI7zqc8rd3PzHXmshjWHJVmZbW2WceUy5VFX9ATVDPl0usT2NHbFH0KAOgK1UtnTlNqxzcpT2vXI6c5c1kMa45Ks7Ja26xjyuXKoi/oCaqZcun1CezobYo+BQB0hfFLZ+75olLPX0B5Wjvu+aIzl8Ww5qg0K6u1zTqmXK4s+oKeoJopl16fwI7epuhTAEDXGL90ZvR8pZ6/kPKsdo+e79xlMaw5Ko3Kem2zjikXK8u+oCeoJJX1MRzx0dvdW/QpAKCrjF868/MzlXphHuVZ7fr5mc5dFsOas9ec1SvUnNUrWt7OyM8vU3NWr1BjW7/bscfMorJe23lYx4PD/eNzPDjcn/n8Dw73q7nrlma+X9o9fr3fbfu83ZVlX+ShJ4Klj5fzf7I4le3l+Zja6cr6GI748tjbZs3/yWLVf9cP2v44jf7m2sr1YwZ9CgDoKtVLZ76k1AslyrPacc+XnLssxtU1N/8ni8dfqJrVqcdP6/GGHrpSzVm9Qo38/PKOPWYWlfXajruO565bWrOexrZe0tF9rh+v/GYt2/nvv+sHoY8X1n9x13LW4zerNiDt7Biz7Itmju36eGWrsa2XtH1/jfz8ciMgbX17negp3dNR5UrPuLpWkYyrr9v0MXnO6hVq7rqlanC439rLc9ctTa3H4/Zn3OO/66/D6FMAQNcZv3Rm+4VKvTCf8qR2b7/Q2ctiXF5zOqjJehzN1tBDV1XeIH8/87G0q1xZ21HreGzrpWrO6hWq/67ram6vhj+XdmRf6XEMDl+T+bz133VdZG8NDl9Tt3b1bZ3aX62MP+tyoS+aObbrY9bQQ1fV3ebzcazZCvb0/J8srpzZnJ/95sJaRTKuvW4LHo91X+heMKsakHa+P/Nc9CkAoCuNXzrz6FlK7fwO5UntevQsZy+LcXnNzR+6ofyiO+b9dA399OqG9xnctGj8//ffdZ1SO79Tf4bhtsusjze27TLrmUL6ccN+Pn52grGtuI858ugVNfebP3SDGvrp1Wr+0A2s7QbreP7QDeU3aoHbh356dc181Jz1Fdjf5pqyniX26BV129dvGoM1uGlR4vnXYw1bR+b6Ntd2cH3YtmO7r96G+bz0Pgn2V5z+s+0323O17TNz7uKOP3g/3eO28cwfuqFu/wXva+7T4D62rS0X+qKZY7veb2FzHDVXYcei4PoYefQKNXfd0vG1FTxe2vZn3PXRyjE1uJ0460OPXd9mHmv0vtTPM+7xxfb3I87xpdm/BS6sVSTj2us22zFW901U79p6Kni8CDsmJzm26N8PjkHfr9Exw+zX4N8CW28G+9IcZ9jfR/oUAACL6qUzX1Zq53cpT2rHPV929rIYl9dc9U15+H2qb2zL/9YvZMsv1su36Re/wftUX9zWblO/AB7bdnndz8a2XW684K69rfzCt3rf2jfI0c816jH1dszH1M/BfO7dvLbD1rFtvqLKDNb07wxuWhS5n21zp9euvk2PI2wsUfOv39SZv6dvM9eWfkzzfnPXLa3pheDvN3pOtu2bt8XpP9ttevtmz9jG1H/XddZ932j8ZgUfO/gz8znZetm2r6Pm0pW+aObYXhuQhh/L9DHV3CfB42zYbXodBR9D/yys15Ksj0Y9ZVt/wZ6Nsz6Cxxdz/Lbjf6Pji/55OViq/Z1gX8fZ/3HKhbWKZFx73abXXpzXI3PXLa1Z342q9j/MJTu2mP1pBpu2Y0KjY4b5ujHYm8G/8+bzMx+3lX1MnwIOO/vq+zbPXfTA4xRFhdfZV9+3udkeG7905rmLlPrl96ic1+7nLnL+shhX19x4QBry8/4N11t/Prjp2sqbxyvV0MP94/8/7D5hvz+2bUHdz0Yerf2cvv4N14eOL+yxbRX1mOMvuC37x3Z7t65t2zoe27ag8gbp2ljPSc9Z3Puba2Lo4f6abeh/x9l22Pzr8dvW2dx1y2rm39Yv5cBmWezeCY4nWOa24vSfbZxh4w2eDWTrhbjjD/ZPWJ/aesu8f9j60XNum0uX+iLpsb3R2h16uD/0PsF9FXaM1fvOdlwMW69J10ecnrLNXXA9NFofScbf6Bige6VRv8bd/3HKpbWKZFx73Wb7e2Hr0bC/B422bfZc3GOL7gnb368kxwxz22Z/Bf/2h/1davRalj4Fcu7MhcNq+IldFEVF1JkLh1WzPTZ+6cxj5yr1y8uonNeux851/rIYV9fc/KEllReV9tvmDy2pvOCt/b2RRyufo/fwNWpw0w8qL3ivqLnP0MOVz1h89Kq63w/7neDP698MxH+MuI85tu2KyovyH2Q+H66vbds6tu0/vT7MSjJnYfM/9PA1Tc9l2O/Yxlr/hi9ev5jVv2Gx9fbgeMz9oMevHzNO/6lfXlZ5c7s41nO2PUfb7zYav1lh29A/mz+0JPS24HOJM5cu9UXSY7te/8Hna/ZF2PoPzleSOTLXq21NJV0fcXoq+Bxtj99ofSQdf6PjS9R+1c8l7v6PUy6tVSTj6us29cvL1Nx1y+r+tpo/C+sfsz+j1nXcY4s+Tutq9Doq6rWfHpfttYS+Laz/mzkW0qdAjpy5cFhtf+UPFEVFVIsBaeXSmROVGrucynntuOdE5y+LcXXNjYc7IbeNv+gM/N7Qw5XLgx+7evz/Dz28qOY++ndHHru67vcH76u8SH56Yaxx6vvPH1oSOo6427A95vibg8DtY08vtD7/bl3bYet4/A1ZjH3eaM70ujHX08hjV9fcFrYNfb/B+34Qe/71HNt+J2m/2J5Ho/UYfA7mOOP0X9T+bzQGvfZt44jzu436R/8sODbzNr3/g7+vn6NtXlzqi6TH9rDjpZ73OD3SaE3r+bPtu/GAIcZYo9ZHsz0VXKuN1kfS8Tfad2G9kmQbScqltYpkXHndpv+uBfssrP+i1rhe28HjbfC2uMcWs9f138KwvwWNtmv7+x28Tf9dsj3nVl6r0aeA4whIKapxVQLSaSLS00yfjV86s+MSpcYWUDmt3Tsuyc1lMS6uuWq4Y79t7OmrjDerC0Jvq744Lf975LF+4411f93jDt53XeVF7lWhPwveXn3hvaDucYYevrZm/OU30Mkf09yO7Xmxtu3reOjhaytvYq6rm7PgPtf3ta2LsPnT2zHnZ/xSWct8BcfRaP6roWx1+3qdm4/RqF8aPZ65hvXPg/uhGkzG7z/b/rfdZts3Yful0fgb9ab5M3Osttuq/zGlv+Y+tvG62BdJju16Xsy1po9j5nPVa9nWO+a+Dq7RqH4M669m1kfcnjLnNGyeG62PuOM3n3vY8cW2D237O+7+jyoX1yqSceF1m153wZ6w/Q209Yj5OsnWs7XBZm0fNjq26L9H+t+6x8N6NOqYYTsOBm+z/f3T22z2tRp9CuRAVEB6zHFK9fYq9e3vvlVz+/2P/kH19qrxajV8MrdlqyTbWj30x1i/9+3vvqV6e5U68qjG2w8+395epY45rvXn3Yk68qjyeE8/663InwfnOO6+DVsjYWspbBxhtWjpH2vmKO78pl2VgLRYqcSql858Q6mxhVROa9dj38jNZTEurbn5Q7XfKGor8/7m5VzlF6vXx97myGPXKDW2UA3ed33k4zW6n+0xw+4f9TPbY6qxhWro4frLKssv8lnbpqh1bNvHY09fHbp/dc0fWhq53kYeu8Y6J8E1F3yMVuffHH/wscaevtp6W1RP6OfZqP+C24nTf2NP13/7vPn8wuYn2Otxxh+1LfN+tl4K66+4fe9iX8Q9tkf1QHCu1NhC1b+h/tumbcek4PrQvRD3eJ9kfbTaU2HrsdH6sD3HZo8vtl4J7rMk+z9vx3DE58LrNv33z/w72GgtRh1Hg+s6uF3z70/UscU8Fpjbt/29anTMCI6hf8MN1tua/VtGnwI510xAmmY4GtyeawGpvl9Y3f9o50PPJGWOP2pfxXkenQ5IbXOU44C0fOnMvbOV+uerWqr+DUtqXzxsvi7y/uN/yJt4rJHHq9/WOH9oaazfmT+0VPVvWNL08zPfXDV6bp2uHffOzs1lMWmuuTyUflE89sw1mY8lj+Xq2u62dUxlU/pvXfBvjot9QU9QtnJxrSIZejsfNX9oaTkwpk8BPyUNSPUZh2mGg3p7q4f+2PK20gxIzW2Z+8C83fUzSc2zX4P7N8lZtEn2bdRaajUgzapaDUhFjEtnnr9cqX++uuUaefxa4w1d+P2qAWnzj1UNSBvfd+665bHvG1ZDm6+L9dw6Wbufvzx3l8WkveZcrvlDS1ted91arq/tblrHVDY1eN/iur+TLvcFPUGZ5fJaRTL0tvtVPlN1MX0K+CpJQNqOcHT7K8kC0kVLqyGdLaAMC/FOP6t6JuWRR6lY4Zv+HVuo9+3vvlXz2MEwNbifbJfpRwWQYfs5zn2CFXaZve3yej3n5r6K2rdhZ5Caa+X0s96yBqS257Jo6R9rwlGzvv3dtyJDWvMxbevJvN28b5yQO52AtHLpzOPnK/XiNS3XyOM/qISI16eyvagaD0hj3Hc8IG3h8YY2X9+x5xa3dj1+fu4ui0l7zblUc9ctr7+s0YFx5bFcX9s+r2Mqm+q/e0n95ZKB+7jcF/QEZZbLaxXJ0Ntu1djotZaPEFhMnwI+ixuQmsFZGmd6NhOQmiFnWNhoC9Cifi8qINUhWpznawv79LaDoa4txLMFqMHnEec+tmp0qboOWMP2kw414wakwaDStq2o53L/o8kC0qhtmeMKu48t4G1TQFq5dOYrSr14bcs18rgOERdbfx4MkMZGr7Peb/5Q8PN7qp8X1H/3UqVe1B/YvqzmvoP33VCznf67wz93LHjf2mCrdrtDmxeroc2LQz7ba5lSL14b+vM4z7fZ2nHvV3J3WUzaa47ys1xf26xjKotyuS/oCcosl9cqkqG3/S36FMiJOAFpklCxlYDUVvqMQjMY07eZvxsV4kWFZVHPJUkgHHY5vm2M5r4NhqjmeHTQqJ9vnPs0Gpt+LjqANM+eDO63YLAaJyC1PZZ5m94Hetu2OQmeRRr3LFbzfrazeG1zrn+30VmkaQSkIsalMy8sVOrFH7RU1YD0hsj7Dd63pBIYXl/3Mx2iBrc5Z/UKNfJ49f7B28ZGwx+7fAbpMutY9O+ZP9fjM8cxtPkG6zjM8eqxlkPcH9RsQ4eure5jXbtfWJjby2LSXHOUf5WXtc06pjpZeegLeoJycK3OFpHhBjW7i7cTC73tXznWpwCiJAlIzdApKpBrR0BqC/RsoWEwQAu7JDvOJfbNBqTmJe9hj2+e9Xj/o/VnoNoCuzj3Cavg5fRR314fPCMzSUAaNk9Rn0EaPHM1aUAaNk/B8Ni2dsPG276AtHLpzC8uUOql61uqkV+UQ8Shny2JvN/g/ZWAdPvimtuHfla+feQXN1jvb94+Z/UKNX/9spr7lcPJZXWPN3fd8rr76uq/u3ymaPD2+etrb9djG7y/9rmNbV88frt+/nqc5v3nr1+m5q5b3vI+1rXrFxfk9rKYNNcc5V/lZW2zjqlOVh76gp6gHFyrJRFRDarUxduJhd72rxzrUwBR4gakOnwyw6ykwWdYqBoniAz7PNBGgWhYQGk7GzNsXHHC4GYeJ+yszmCZ+yXOfWxlho3BcNY232kHpMH5s30kQTMBadhzsY0tan82Ois6xYC0cunMSUq9dENLNfKLJZWAdGnk/QbvX1oJSJfEun3oZ0srwWP19nJAurzmfrbb1Es3VALS+tvVSzeo+euXV4LL2turwWntGGzPTQez+vmbt+sxhz1Os7Xj3pNye1lMmmuO8q/ysrZZx1QnKw99QU9QDq7VkrgVSLq2nVjobf/KsT4FECXpt9ibgVSjbyRPMyDN4gzSqLMeFy39Y8MzG6Nujwr2gkFo2NmNce5jG4cOK83fMT8nNRjYtuMMUn12p+0S/3afQZplQCpiXDqz8xqlXl7SdI08YYSIEfcbvH9ZJQitvV9NCGnc3n93+TL2kSeqt4+Hocb9bLepl5dUA1LLWPS2g2MZv3S+wdj04/bfvVyNPLG05nHM3x8PSFvYv7p277wm95fFpLXmKL8qb2ubdUx1ovLUF/REd5eDa7VQGcPnROSUkPorEZkiInt14XZio7f9KQf7FECUpAGpeTZgnDMX41ScbWXxGaRhnytq3q4DvqhvV7cFysHPzbSFlsHwLs59oir45Um2S81tn8/Zjs8gDY7BNr+2s2/T+AzS7ANSfenMhUq9vKzpGnliWSVEjL5fNSCt/9ncdZXP7QxssxyQVu9XDUOjb1MvL6uEk/XbHPrZMjW2vf739PjMcQz9rHqbOe7565eP3zbyRPl3zZ+FjaGV2vWLC3N/WUxaa47yq/K2tlnHVCcqT31BT3R3ObxWC1IdT1gVung7DdHb/pTDfQrAppmANG7AmGZAGgxmgxUVoIVdlh5n/FG/a4ZvUQFpq99ir/d/3G9rj/tcwj4rNWwfxf0W+7BL9c2ANOo+wYDUfI5pfIu9AwFp5dKZk5X61YpENfJE7TfT131b/APLlfpV+X+j7jfyxPLxbc5fH30f87a561aE3mZW2LjCfj4ekAbG03+3/bGGRpbXbbscvNZvI+k+DtaOe0/O/WUxraw5qnGltdY6XXlb26xjqhOVp76gJ7q7HF6rBXErkHRtOw21o7fN1876NXbWazhYcV9P2d67jD3b3La6uE8B2DQbkKb5hU1xA1JbaBa8lDssqDTD1SOPUonOvLQFcHEfN2obUfsiKvhsJhwNjtF2Sb45p3q/mGFq3IB0+yu1Z6ueftZb1o8rCIai+vFtZ9rq26P2c/AM2eB6ciUgFTEunfnl9Ur9aqVTNfiAfqHRucecv16Hn9k/f7N2//J6by6LaXXNjTwxEBpsl4Pp7Ocrq6q+CM9+LL6v7XYfO/Xxz/4foPI1x1TyymNfJOkJ1rc/5fha3UNEJjSoPbp4O7Ek/XsXDA3167aRJwbq7lsNSLNfy8FxNXo9pU+eSGNbXd6nAGyiAlKKosqVfkBauXTmiZJSvxp0quav14FX5x6z+uYs++dv1q4nSt5cFpPWmqsGotG3UW5XXtd2J46d+k2leUyy3Ub5V3nsi6Q9wfr2o/K4VpFMkt6uhoHV2/THaA2NDFjv3393/e15qLnr9EkV2Y+FPgU8REBKUY2rDQFp+dKZ4VOUeuWHmZZ+AVU9G3CgrY839twqy5kr7X3MZmvH8CneXBaT1pqzrZH+jeU317bbotbVyJMr69ZBzeVSz62quW3++oGaf89dV/uYwbUctq6C95uzeoXq31h736GRlXX30WMK2475M9t+C7vv/PUD1n3Qzl7M69ruxLFTr8vg+pm7rn5O4qy5OOs8uAbMf5tr03b8HHlyZd1j2u5XfqNce99gn9r6KrhGbduJO37XK499kbQnWN9+rO88rlUkE7e3dc8FX4Po3rT1UNTatfVWu/swzusp23Zsx6a4r81sr/Oi9kXwWBd8nvQp4AkCUopqXGkHpCLGpTNjS5V65SbKsdo9ttS7y2LSWHPV4K7876GRwbrb9IvTkSdX1d0W/L3BB1aO36bfdJrbCv6+uY3yC9RBNfbcjXU/07cFt1UNAWpv69+4sm5s5vj1bWPP3Vg3tsEHVob+bOTJVZU3Iivr7j80Mlj33M39ERwXa7us3cdOPWfmXPRvXFlzW9w1l2Sdm9vUP9NjGXvuRmuv2bavt2O7zVxz+jmZY+jfuLLyxrW2583bbGs6zvjbMVf0RVmSnmB9539953mtIpk4vW17XdOobGs8rGyvcdLuw6jHsj3fsN+Pu63gMc+8zXztZx6HzPsPPhD+HOhTIOcISCmqcbUnIK1cOvPkxUq9ejPlWO168mLvLotJY83Z/8v9yvGfD20pv2Ec2rKq5vfGdtxUc9+568pn4di2b7t97rrKWaqWMekXtWM7bqq5XY9l5MlV1vHPXVf/OOrVm8dfVI+fJRB4LmH3Dz5+1PPUIULYv/VtYWPs5rXd7mOnDj3qz6CproO4ay7JOtc9Yru/3pZtPYy/AQ0Zf//GldbtBc+as63f4LaDa97cJ43G73rluS+S9ATrO//rO89rFcnE6e2wPoyqqN4JO2aYr4XS7ENb70XdL+xYEXdbupdtY7btS33cM1/r0qeAx85cOKzmD2x5vbTs4X+nKMpebQpIK5fOfFWpV2+hHKsdw1/17rKYNNZc+Q3iSqVevUWN7agGjvrngw8M1rwYDlb/xsHx7ej/b1b5xenK2LerV29R89fbL4evBpw/HL/vyJM3NrxP2DbLL8jrH18/57EdN1v3l+15Bn9HP55tDKztWu0+duo1MvhAed6GtuiPlqiuv7hrLsk61/2kH9e2lqLKXH9hfRhco1E9qp+nbd3rfWT2TaPxu1557oskPcH6zv/6zvNaRTJxejvq9VFYhfVuVH8FXyel1Ye2bdpeT5nPN+z1WJxthb0ODHutp4+RI0/eSJ8C3eDMK4eLZ145XDzic397NkVR4SUpB6QixqUz/7xSqVdXU47U7n9e6e1lMa2uueob6fK/R568yXhBuVoNbbmx8kLypsjtVF/gVm8b23FL3fbN+9tuV6+uVv0bBytBYvLno8ff6Pd1YDD4wKq6nw0+ULnMcsctsccdHHM1DK1/TNZ2vXYeO/WaMOdaz3H5TV38NZdknevbbWusUQ80Kj3+Rr+ve0H3b9jztPV5o/G7XD70RdyeYH3ne337sFaRTKPejurXuesG1NCWG629oPvdti3zd/Qxw7adNPrQtg3b66mo407UeILbStrLcV/X0qeAf4oURcWq1IxfOvPUJUq9toZypHY9dYm3l8W0uubGX/watw0+WLn86MFVSr22Rs39p8oXBjx/6/h99AvM/o2rav6tf0e9tmY8EAxuX2/Tdrt6bY0ae/7Wckj7TwM1t/dvXFV9Uf/aGjXyVPkMoJGnbq653/z1K2t+d/76lXWPpX9Xb8v2/M3nG3ze5vOMeu7BcQVvY22XtfPYqefanB9zPkaeujn2mkuyzvU2g48b1kNmb5jrT69HW++av2t7rOBa1mMyx2q7Lc74XS4f+iJuT7C+872+fVirSCZOb4//h15LvwZf89j6xezfYN9XzyCtvv5Jsw/j/sw8JgTHaKuobQWPZWaPB7c9HpBa9iN9CvitKOX/skFRVHgVJUXVS2dOVeq1tZQjtWP4VG8vi2l2zY089Y91lyENbblp/Of6TWz5heXa8RefNZctVX4Wtc3qG9Pa7dpq5Kl/rNme7b7lF+rxH0+9tjb08lLz8QYftH/Tq+2+Y8+vjvx58PHGnl9tvY21XdWuY6dtXvW+N+dRr6tGay7OurOtD7PMbYXd1+zFsLU5+OAPa7YV5/HCesvcVpLxu1o+9EWcnmB95399+7BWkUzcv3e2113ma4eodRz1usvscf06Ls0+jPt6Ks7rwSSvzYa23GS9j95nYT8P7i/6FPDXBIqiYlWqxi+deelGpf5lHZVx7X7pRu8vi3F5zc1ZvUL13/PDzMfhY/m2tl1ex42Kde5O+dQXrvQE65u1inS50tsUfQoAQFtVL525TKl/+TGVce166jLvL4txdc2NbC2fqTOydXXmY/GxfFvbrq5j1nm+yqe+cKEnWN+sVaTPhd6m6FMAANqueunMHKX+9XYq49oxPMf7y2JcWXNDj9R/u+nI1tWZrwFfy7e17co6Zp3nu3zqiyx6gvXNWkX75eXvHUWfAgDQsvFLZ16+Ral/HaIyqt0v39I1l8Ww5rqrfF3brGOqlfKxL+gJP8vHtYpk6G33iz4FACAF45fObF2g1L/eQWVUu7Yu6JrLYlhz3VW+rm3WMdVK+dgX9ISf5eNaRTL0tvtFnwIAkILqpTNfU2rXnVRGtWP4a11zWQxrrrvK17XNOqZaKR/7gp7ws3xcq0iG3na/6FMAAFJSvXRmrVK7NlAdrt0vr+26y2JYc91Rvq9t1jHVTPncF/SEX+XzWkUy9La7RZ8CAJCi8Utntl2p1K83Uh2uXduu7LrLYlhz3VG+r23WMdVM+dwX9IRf5fNaRTL0trtFnwIAkKLxS2c2fU2pX99Ddbh2bOq+y2JYc91Rvq9t1jHVTPncF/SEX+XzWkUy9La7RZ8CAJCy8UtnfvVjpX49THWodv/qx117WQxrzu/qlrXNOqaSVDf0BT3hR3XDWkUy9LZ7RZ8CANAG1Utn+pV6/T6qQ7VrW3/XXhbDmvO7umVts46pJNUNfUFP+FHdsFaRDL3tXtGnAAC0QfXSmdOVev0BqkO1Y9PpXXtZDGvO7+qWtc06ppJUN/QFPeFHdcNaRTL0tntFnwIA0Cbjl868eodSbzxItbl2v3pH118Ww5rzs7ptbbOOqTjVTX1BT+S7ummtIhl6252iTwEAaKPxS2eeXqTUGz+l2ly7nl7U9ZfFsOb8rG5b26xjKk51U1/QE/mublqrSIbedqfoUwAA2qh66cwZSv3bZqrNtWPTGV1/WQxrzs/qtrXNOqbiVDf1BT2R7+qmtYpk6G13ij4FAKDN9KUzv3nhRvXmP/+IalP95oUbuSymgjXnV3Xr2mYdU1HVjX1BT+SzunGtIhl6O/uiTwEA6AB96UyeatX3Z6mBS4tq4NKinkNGgwAAIABJREFUeuL2v8l8PEmKy2LyueZYw6ztINax/2s6jeqmvtjmcU90w/ruprWKZHzu7bz1OX0KAEAbPX3H3xQf/6e/fuj/sHf3Mbak9X3gf7JHK2RFybC7RBCtdo+0soRWm/XIOAiNnPissLQmRvEoC3jyZtrh1ZhdJuYmsuUX2tfBBoIHuFxehrnD9FwMNhnWGgd6MDBwBwbMuLGH8cDcaxMuc31t85YMdhKiKDeG2j+6q091nao6Vd2nu6qe+nykr2BOnz7nOVW/p/qp363q/tRv/L2Hx5LvesJ134qILCKy7duf9bm+x9MlP/sTN9wSE78tZow1p4bVdpk6Tr+m15EpzYuU58QU6ntKtUo3Kc/tsc1z8xQATsb3x+IH7tDzzdhbwETEswcwnsPEbTHjqrl1J4UarsvUanvKdVxMyjW9jkxpXqQ4J6ZU31OqVbpJcW4XM6Z5bp4CwDF6avT/w75tiguYH43hL2Kq4raYcdXcupNCDddlarU95TouJuWaXkemNC9SnBNTqu8p1SrdpDi3ixnTPDdPAeAYXR8RN4wk/zUWC5iXRsQLI+JpAxhXl1zfbrckbUw1p4bVdp0p1/FUanoduT6m4/rof3urb7XK+qU4t8c6z81TACAiFouXLBb/knpdj+OBrtQwqVHTpEx9Q/rMcwBgdCxgGDs1TGrUNClT35A+8xwAGJ2qBcx39Dge6EoNkxo1TcrUN6TPPAcARqdqAQNjooZJjZomZeob0meeAwCjYwHD2KlhUqOmSZn6hvSZ5wDAqDwhFouXa2EBw/ioYVKjpkmZ+ob0mecAwOhcH4sFzDfDAobxUcOkRk2TMvUN6TPPAYDRsYBh7NQwqVHTpEx9Q/rMcwBgdKoWMH+nzwFBR2qY1KhpUqa+IX3mOQAwOlULmO/vc0DQkRomNWqalKlvSJ95DgCMjgUMY6eGSY2aJmXqG9JnngMAo/PkWCxgvhEWMIyPGiY1apqUqW9In3kOAIzOLBYLmK+GBQzjMws1TFpmoaZJ1yzUN6RuFuY5ADAys7CAYdxmoYZJyyzUNOmahfqG1M3CPAcARmYWywuYG3scD3Q1CzVMWmahpknXLNQ3pG4W5jkAMDKzWF7APKPH8UBXs1DDpGUWapp0zUJ9Q+pmYZ4DACMzCwsYxm0Wapi0zEJNk65ZqG9I3SzMcwBgZGZhAcO4zUINk5ZZqGnSNQv1DambhXkOAIzMU2OxgLkaFjCMjxomNWqalKlvSJ95DgCMzg2xWMB8MSxgGB81TGrUNClT35A+8xwAGB0LGMZODZMaNU3K1DekzzwHAEanagHz9D4HBB2pYVKjpkmZ+ob0mecAwOhULWC+r88BQUdqmNSoaVKmviF95jkAMDoWMIydGiY1apqUqW9In3kOAIzOM2KxgLkYFjCMjxomNWqalKlvSJ95DgCMzjwWC5iHwwKG8ZmHGiYt81DTpGse6htSNw/zHAAYmXlYwDBu81DDpGUeapp0zUN9Q+rmYZ4DACMzj+UFzPf2OB7oah5qmLTMQ02Trnmob0jdPMxzAGBk5rG8gLmhx/FAV/NQw6RlHmqadM1DfUPq5mGeAwAjMw8LGMZtHmqYtMxDTZOueahvSN08zHMAYKBuiohXRcT1pcd/KBYLmJ1YXsDMImLz2EcHq6lhUqOmSZn6hvSZ5wDAKF2JiD+PiFcUHrspFguYT8ZiATOLiDv3Ht84uSFCoyuhhknLlVDTpOtKqG9I3ZUwzwGAkdmIxWLlsYh4fiwvYG6OiN8qPHalh3FCnY1Qw6RlI9Q06doI9Q2p2wjzHAAYoSuxWJxkEfG1mv+fZ6OPQUKDK6GGScuVUNOk60qob0jdlTDPAYCR2YjlRUpdrvQyQmi2EWqYtGyEmiZdG6G+IXUbYZ4DACP0F9FuAXNTXwOEFdQwqVHTpEx9Q/rMcwBgdDZj9eLl/p7GBm1shhomLZuhpknXZqhvSN1mmOcAwMhcH6v/lXfe1+CgBTVMatQ0KVPfkD7zHAAYpc2oX7w82N+woLXNUMOkZTPUNOnaDPUNqdsM8xwAGJmmf+Wd9zcsaE0Nkxo1TcrUN6TPPAcARmkz/G4gxm0z1DBp2Qw1Tbo2Q31D6jbDPAcARqbqX3nnfQ4IOlLDpEZNkzL1DekzzwGAUdoM/7rLuG2GGiYtm6GmSddmqG9I3WaY5wDAyLwsIv5b7C5gPhsRF0RGlgdCDUtaUdOSctS3SPppmucbAQAwQJtR/9cmRURERERE1pXNAAAYoM3of6EkIiIiIiLpZzMAAAZoM/pfKImIiIiISPp5e0TcEBFPDgCAAdmM/hdKIiIiIiKSfrZi96/azwIAYEBmsbtIuTkibhERERlYqk6wf2YvfY9NRES65ebQIAUABmwWu4sVERGRIaWqQZqfaPc9NhEROVxmAQAwQLPof6EkIiJSjgapiEh6mQUAwAA9ISKuFxERGViqGqTP3kvfYxMRkcPlCQEAAAC0UtUgne8FAAAAACBpGqQAAAAAwGRpkAIAAAAAk6VBCgAAAABMlgYpAAAAADBZGqQAAAAAwGRpkAIAAAAAk6VBCgAAAABMlgYpAAAAADBZGqQAAAAAwGRpkAIAAAAAk6VBCgAAAABMlgYpAAAAADBZGqQAAAAAwGRpkAIAAAAAk6VBCgAAAABMlgYpAAAAADBZGqQAAAAAwGRpkAIAAAAAk6VBCgAAAABMlgYpAAAAAJCkWYvnHKZB2uZ1AQAAAAB6NYuIKxHx/IbndGmQbkTEY6FBCgAAAACMxFbsNj0fi+pGaZsG6cbe92d7rwcAAAAAMAqzONj8LDdKmxqkG7FojOaZncCYAQAAAADWZiuWm6B5o7SqQfqaiPhKxeNbJztsAAAAAICjm0V1I7RrZic7bAAAAACA9diKozVHt056wAAAAAAA6zILV48CAAAAABO2Fa4eBQAAAAAmahauHgUAAAAAJmwrXD0KAAAAAEzULFw9CgAAAABM2Fa4ehQAAAAAmKhZuHoUAAAAAJiwrXD1KAAAAAAwUbNw9SgAAAAAMGFb4epRAAAAAGCiZuHqUQAAAABgwrbC1aMAAAAAwETNwtWjAAAAAMCEbYWrRwEAAGBaPvjSmL3/hdfNRUSmnn/2t+PmiMj+2d+Om/sei4jIkPLBl7qqHgCAhL3/hddtfuBF12VTznbpf0VkuvmVH/7O3scgIsOI9cEi73/hdZt9r1kBAODY5A3Sj/7Uk7NP/eLfnGTu+6knZx940XXZfRPeBiIiInIw1gd/M/voK5+iQQoAQPryBumlO27Orl163yRz6Y6bs6lvAxERETkY64P3ZZfeubsN3vtj170mIq6PiCf0vXYFAIC10yB1AiQiIiLLsT5YNEjfcNN3bkXEPMLvIgUAIEEapE6AREREZDnWBxqkAABMhAapEyARERFZjvWBBikAABOhQeoESERERJZjfaBBCgDARGiQOgESERGR5VgfaJACADARGqROgERERGQ51gcapAAATIQGqRMgERERWY71gQYpAAAToUHqBEhERESWY32gQQoAwERokDoBEhERkeVYH2iQAgAwERqkToBERERkOdYHGqQAAEyEBqkTIBEREVmO9YEGKQAAE6FB6gRI0s4rX/AjWUTU5pUv+JHexrZ97uf2x5E/9vTv+e61jCt/3e1zP3fin6vqvc++6sUHtnvxv/sYY1PyffCcZ93Y6nOefdWLex+zyHHE+kCDFACAidAgdQIkaUeDdBgN0vJ2L352DVKRYcb6QIMUAICJ0CB1AiRpJ2+QPv17vrv3sZSTaoO0nEsfOjuo8ayKBqnIbqwPNEgBAJgIDVInQJJ2ujZIn/KkJy5d6Vj8erGBmT+nqZEWpSsni889SoO0fMt6+fNVjb/4flHT3Mvfv+mzrXpO8b2r3vMpT3pi7RjLn6v82l23f3ms+XvX7fPnPOvGygZpscmbb7e6bSiSSqwPNEgBAJgIDVInQJJ22jZIyw2wYorNynLDLRoaZOVma/n5h22Q1v3agOJnzB/Lm49Nn+/Sh85m1y69L3vOs25c+fnbPKf43l0apOXmaNXn6rL968ZabHzW7aPi85q2XdP7i4w91gcapAAATIQGqRMgSTtNv4O0eDVh3ngrPlb83ryJWGzQNb1v3hisau7ljbfDNkjz78mfU2zg5ePM/ztvPuafpaqRmTf48mZh/t95w7K4Tdo8p/zedbfYlx/LX7v4nPJjbbd/8fXL+zMfa9XvQS0+lu+nYh1U1YYGqaQa6wMNUgAAJkKD1AmQpJ22DdKoaOBdu7TcEGz7+ymLqbtt/DAN0qrvqUrd57l2afnKyvJna/reNs8pf61Ng3TVVZr59jjM9i+/dr7f664uLr9H3T4pbz+R1GJ9oEEKAMBEaJA6AZK00+YW+6orMPOUm2Nd/ohS1DT7+miQVt3qHhUNvqqv1/1+07rnlN+7TYO0aXyH3f5Vt+NHiwZp3kBe1SAtN89FUov1gQYpAAAToUHqBEjSTtvfQRodryBd1aArXqWZP7aOW+wP2yDNP0dVI7OuwReFpmLdX5+vek75v7s2SJs+12H+iFX51wy4glSkXawPNEgBAJgIDVInQJJ22jZIu/4O0lUNuqpbwfPXWvfvIC0+ljfrIqqbllV/IOrsq168sonZ9jlV7932d5CWP1fV97Xd/uVmaNU+9jtIRZpjfaBBCgDARGiQOgGStHNcf8W+bYOuKkdtkNa9dtPvVK273TwKDb66v/peHF+b55Tfu22DtO5zVf2hq1Xbv+mW/aqmadN+Ko61afuJpBbrAw1SAAAmQoPUCZCknbYN0jz5rehR0dC7dunwvwPzOc+6cb9pV3UFY9fXL//hp2LT79ql6l8ZEKWmXr5tio3AqiZl3Tate075vds2SKs+V/mPMXXZ/sVx5tsn/+/i75ot7vPnPOvGyqt/yw304jg1SCXVWB9okAIAMBEapE6AREREZDnWBxqkAABMhAapEyARERFZjvWBBikAABOhQeoESERERJZjfaBBCgDARGiQOgESERGR5VgfaJACADARGqROgERERGQ51gcapAAATIQGqRMgERERWY71gQYpAAAToUHqBEhERESWY32gQQoAwERokDoBEhERkeVYH2iQAgAwERqkwz8B2j73c1lE1Gb73M/1PsZiXvmCH8kiInvKk5545Nda12csbsOjjunp3/PdWURkr3zBj6zttZ7zrBtrn/OUJz3xwP4++6oXt3peVS596Oxa9vFznnVj7ZjL9XqYOlj1Wety9lUvXkvdrfqMXXPpQ2cPbP911mPd2FfVQtdtW7V/juvY02ZeHPf+TyFHrbPiMaXLdm1bH+uoxePO0NcHJ7INNEgBAJgCDdLhnwCtapDGwJqkGqTdX2tMDdJVY+6rQbrOujtsg27V9tAgPb7tv879n0qOUmf59szz9O/57rXXxzpq8bgz9PXBiWwDDVIAAKZAg3T4J0BNJ7n54+u60m0d0SDt/lptGqRd3i/fB12aGusc87VLu1fzHbYO4hCNEw3S1e9/3O910jV2XPs/lRxl3+dN9sMcQ9oetw8zz086Q18fnMg20CAFAGAKNEiHfwLUdJJb1QgrX/lT1TDIm1d5yo2IqiZgVaPn2qWDV6s95UlPrG1UlN+zzYl3tDzRLj636jMVx16+wrH82vnt0HXbr65BWr6Cs+pqzeJznvOsG0+8Qbpqvxc/X9Vzyl9r2jdtG6Tl7V0cY7FxUnUldf71cs0Xt1fT91Vl1Wcs7+em2qx676d/z3cvzenia5bHtqoeD3P8KD5W3HZ5zZa3Qfk9y5+9WG/lGmszd9vMi8Pu/1XHhrZ1WT7+lo+Hde9dl/JVvlU1uWo/VL1O3TF7VZ2tmgPF56+aA1WPt53nQ8vQ1wcnsg00SAEAmAIN0uGfAHW5grTcnMhTbJS1eU7bBmnTrbzFk/mm5zV99vw5q5osdbeX5yfebX+Pa1VTpDzOqm1T9/5NjbViTqJB2ma/1+2n/L3X3SBt2t7F/df0vEsfOlvbIFv1fVVjqvuMTa9Vt2/aNEir6qJLPR7m+FE1rnw/1dVAsUbL46za/m3H2mZeHHb/tzk2NNVu05w+yu+GXjXP2u6Hqlotvk7bOmszB86+6sWt50CXOl61L/rO0NcHJ7INNEgBAJgCDdLhnwC1+R2k5YZlVWMhf05+olzVvMsfa9sgjYaT47zhUvy+4olw/lhTczB/Tpur9KoavPlrl6+YK3/u/Hn59qtqVJSbhPl/582UqmZk/lj5CtbyYyfRIG2z3/P/zvdTVZNznbfYFxtbVY+VrxCsqrPyc4rv1+b7qlL1GfPHql6/OCea5m/VLfZVjcd8vG3qscvxo+qx8uvU7Y/i5y7XctU+q6r5prE1zYvD7v82x4am407T/m+7D8spHpObHlu1H6q2XXnOHXaMxTqr2nar5kBTfTTN8yFm6OuDE9kGGqQAAEyBBunwT4BWNUiLzy1fSVU+AV51JU9dE7A8jksfOlt7ZWv5RH7V7d5NDbQonWivSt0t5HVjLY+h6Wq2fPzlbbPqD+I0bYOTusW+635v2u7rbJDW/bqC/P2rGid1tyav+h2UbW5pbvqMdduk3FRumr9VDdJiY7W8PdrUY5fjR9OYymm6tb+8Leq2/ar5e5h5cdj93+bXSzQdK8r/2NR2H64aRzltf+VHm98zfNgxFrdz8fXbzoHy8w4zz4eSoa8PTmQbaJACADAFGqTDPwHq+oc26m67LL9WVdbRIC03xspXtXb5XFFzQl73vHK6Nkibtk1dg7Ruexffr66ZUbdtillHg7Ttfq/bBl2vwKvatk21Wvf7XNv8ioQ2VxA2fV/TuJquxF71GermTVM9lt+3TT0e5vjRNPfq6rlNg7Q8plXbuu28OOz+b3Ns6FK3xc/ddh/Wfea6lJuKdfuha4O0yxiL+6DqH1lWzYG6z7Jqng8xQ18fnMg20CAFAGAKNEiHfwJ0lL9EXDzJfuULfqT1a43pCtLiFWXl8R+2Qdr0fnUN0qYGZt9XkB62hvLvKW6TPq4gzbdB1dVsTQ2yNt/XNK6TuIK06X3b1OOqdGmQFq9ubLo6tO4562qQlrfDYfd/m2NDm+117VLzFaRd5nSbxmab/XDSDdIuc6DumOEK0nFGgxQAgEnQIB3+CVCX5lZVM62u6ZI/p3jiXz6hrbpyMGJYv4O0qZlV1SAtPq/8O0irPnd5m5ZP9qt+v165uTCE30G6ar9X1UHx+/J91+aq12uX1vs7SMv/XVVTVe/X5vuqUvUZD/s7SKuuvGvbuGpTj4c5frT9x426z12uk8M2SNvOi8Pu/zbHhqbjTtvfQbrqeLTqMxcbom33w6ptd5QmbnEO+B2kw14fnMg20CAFAGAKNEiHfwLUpUHa5q8v193iWTwRXnUbaH4i3PS8df4V+7qcfdWLG8fQ5hbd4udpel4+pqqroVZt8+L3NY2zKutqkLbZ7232U/k5dc3rNg3SVfs4b5w0bbtygyzPK1/wI62+rypVn/Ewf8X+2qXl3yFZ/iv2xec2XfncZd6sOn7UvX/T+x1Hg7TtvDjs/m9zbDjscfQozccuf32+aT/Uvc6qK+cP2yA97F+xbzvPh5ihrw9OZBtokAIAMAUapMM/Aep6e3TVyXW5idXmD5ZUNYny/1+8Wq74vKc86Ym1vwuw/J5tfodi+XOUU9VAe86zbtwfa93VVk3bpumP0xTfq+736VVto6rnPOdZN57YLfZd9ntVU6lpv9Q1N9o2SMvbuzjGqiuO88fzcVZd5Vd8vM33raq94jjK+7nN7e/FbfqUJz2xU+NqVT0e5vjRdEwpj7W4LVZdIXiYBmnbeXHY/b/q2NC2Lsvb6ijNx2uXVv/RsDb7oep1iu+77gZp2zlQ9XjbeT60DH19cCLbQIMUAIAp0CB1AiTDzmEapENI2wapiMhQY32gQQoAwERokDoBkmGnfLXWkK+2unZp+QpmDVIRGWusDzRIAQCYCA1SJ0Ay7GiQioj0E+sDDVIAACZCg9QJkIiIiCzH+kCDFACAidAgdQIkIiIiy7E+0CAFAGAiNEidAImIiMhyrA80SAEAmAgNUidAIiIishzrAw1SAAAmQoPUCZCIiIgsx/pAgxQAgInQIHUCJCIiIsuxPtAgBQBgIjRInQCJiIjIcqwPNEgBAJgIDVInQCIiIrIc6wMNUgAAJkKD1AmQiIiILMf6QIMUAICJ0CB1AiQiIiLLsT7QIAUAYCLyBulDb3xm9rWP/NIk89Abnpl94EXXZQ+9YbrbQERERA7G+uCXsofe+EwNUgAA0pc3SKec7Rddl71/73/7HouI9Jdf+eHvXErfYxKR/mJ9sIgGKQAASdt+0XduvP9F193/vh+/7sH3/JPrHhYRmWoiIiun7zGJiAwhm//Xd7wmNEgBAJiAWewufEVEppqlBukAxiQiMqTMAgAAEjaL/hfdIiJ9pqpB+uy99D02EZEhZBYAAJCwJ0fEDSIiE05Vg/SFe+l7bCIiQ8iTAwAAAEhW0y32AAAAAABJ0yAFAAAAACZLgxQAAAAAmCwNUgAAAABgsjRIAQAAAIDJ0iAFAAAAACZLgxQAAAAAmCwNUgAAAABgsjRIAQAAAIDJ0iAFAAAAACZLgxQAAAAAmCwNUgAAAABgsjRIAQAAAIDJ0iAFAAAAACZLgxQAAAAAmCwNUgAAAABgsjRIAQAAAIDJ0iAFAAAAACZLgxQAAAAAmCwNUgAAAABgsjRIAQAAAIDJ0iAFAAAAACZLgxQAAAAASNKsxXMO0yBt87oAAAAAAL2aRcSViHh+w3O6NEg3IuKx0CAFAAAAAEZiK3abno9FdaO0TYN0Y+/7s73XAwAAAAAYhVkcbH6WG6VNDdKNWDRG88xOYMwAAAAAAGuzFctN0LxRWtUgfU1EfKXi8a2THTYAAAAAwNHNoroR2jWzkx02AAAAAMB6bMXRmqNbJz1gAAAAAIB1mYWrRwEAAACACdsKV48CAAAAABM1C1ePAgAAAAATthWuHgUAAAAAJmoWrh4FAAAAACZsK1w9CgAAAABM1CxcPQoAAAAATNhWuHoUAAAAAJioWbh6FAAAAACYsK1w9SgAAAAAMFGzcPUoAAAAADBhW+HqUQAAAABgombh6lEAAACOwUZEXBARERlBvhK7zdGvDGAsIiJyMBsBADBSm9H814FFRERERERWZTMAAEZqM/pfTImIiIiIyLizFRHz8CtQAIAR2oz+F1MiIiIiIpJGXh8AACOzGf0vokREREREJI28MQAARmYWu7fC/OdYLGp+MSJ+JiJuERlR/kuoYUkvp2JR1/8tdutabUsKccwWSSv3xcEG6fUR8YQAABiZ/xqLRc2p2F3oPDN2m6ciY8i1UMOSXp4di7r+L7E4Ee17XCJHjWO2SFp5Xyzm9F17j80CAGBkirfF5CfgTlRkTFHDkmI0SCXVOGaLpBUNUgAgCcUTlWfv5X+I3dtjRMYQNSwp5n+JRV1/M3br+u8OYFwiR41jtkhaeVss5vTbYrdB+j8HAMDIFE9U5nv5jh7HA12pYVJ0fRxskM4j4vv7HBCsiWM2pOWNsZjTZ2N3Tv9PfQ4IAOAwqk5UYEzUMCnSICVVjtmQFg1SACAJTlQYOzVMip4ci7r+RmiQkg7HbEjL22Mxp28NDVIAYISeEIsFzbVwosL4qGFSNYtFbX81NEhJg2M2pGcrFvP6NaFBCgCMUNUtnPMexwNdqWFSNQsNUtLjmA3p2QoNUgBg5JyoMHZqmFTNYrlBemOP44F1cMyG9GzFcoP0b/Q4HgCAzqpOVP5OnwOCjtQwqZrFcoP0GT2OB9bBMRvSsxXLDdIn9zgeAIDO/JVkxk4Nk6pZaJCSHsdsSM9WaJACACPnRIWxU8OkahYapKTHMRvSsxUapADAyD05Fguab4QTFcZHDZOqp8aitq+GBilpcMyG9PxGLOb16dAgBQBGaBb+SjLjNgs1TJpuiEVtfzE0SEnDLByzITX3xGJe/1xokAIAIzQLJyqM2yzUMGnSICVFs3DMhtRokAIAozeL5ROVG3scD3Q1CzVMmqoapE/vc0CwBrNwzIbUVDVI/3qfAwIA6GoW/ggI4zYLNUyaqhqk39fngGANZuGYDampapD+j30OCACgq1k4UWHcZqGGSZMGKSmahWM2pEaDFAAYvVk4UWHcZqGGSdMzYlHbF0ODlDTMwjEbUvPbsZjX/yI0SAGAEXpqLBY0V8OJCuOjhknVPBa1/XBokJIGx2xIz/2xmNe3hAYpADBC/koyY6eGSdU8NEhJj2M2pOf+0CAFAEbOiQpjp4ZJ1TyWG6Tf2+N4YB0csyE998dyg/S/73E8AACdVZ2oPL3PAUFHaphUzWO5QXpDj+OBdXDMhvTcH8sN0ut7HA8AQGf+SjJjp4ZJ1Tw0SEmPYzak5/7QIAUARs6JCmOnhknVD8WitndCg5Q0OGZDeh6Mxbx+WWiQAgAj9IxYLGguhhMVxkcNk6qbYlHbnwwNUtLgmA3peTgW8/qFoUEKAIzQPPyVZMZtHmqYNGmQkqJ5OGZDajRIAYDRm4cTFcZtHmqYNFU1SP+PPgcEazAPx2xITVWD9K/2OSAAgK7msXyi8r09jge6mocaJk1VDdL/vc8BwRrMwzEbUlPVIP0rfQ4IAKCrefgryYzbPNQwadIgJUXzcMyG1GiQAgCjNw8nKozbPNQwabo5FrX9sdAgJQ3zcMyG1PxhLOb1j4UGKQAwYDdFxKti+Rem/1AsFjQ7sXyiMouIzWMfHaymhknVTRHx/IrHN2JR278dyw3S2d5zYIgcsyE9b4zdOVp2JRbz+uZYbpBu1HwfAEAvrkTEn0fEKwqP1f2V5FlE3Ln3+MbJDREaXQk1THquj4i/iIjH4mCjdCOqG6SzWNT2/MRGCd1dCcdsSMlm7M7RO+Ngw/NKVDdIN2L3Z9v9JzVAAIA2NmKxeMkqKb17AAAgAElEQVRPxMsnKjdHxG8VHrvSwzihzkaoYdK0Gcu1vREHG6S3RMRvFh67/+SHCZ1shGM2pCT/B718vuaN0iuFx/5pRLy59Nj8pAcKALDKlVgsVrKI+FrN/8+z0ccgocGVUMOkp3zSmUXEvyv8/8djubbnfQwUOroSjtmQks1Ynrf/qfD/v1n62v19DBIAYJWNWF7U1OVKLyOEZhuhhknTZrSv7ft7GSF0txGO2ZCSqn/Qa8q8l1ECALTQdlFzU18DhBXUMCnqctI572eIcCiO2ZCWzfCPeQBAAjbDgoZx2ww1TJo2Q22Tns1Q15CStv+gN+9pfAAArbRZ1Mz7Ghy0oIZJldomReoa0rMZzXP6E72NDACgg82oX9A82N+woLXNUMOkaTPqa/vT/Q0LjmQzHLMhJav+4eMH+xsaAEB7TYuaeX/DgtbUMKm6PlxlR3ocsyE9bwy/MgMASMBmWNAwbpuhhknTVizX9j19DgjWYDMcsyEls6hukN7Q45gAADqruppj3ueAoCM1TKpmsXzCOetxPLAOjtmQnq04OKe3+hwMAMBhbYarOBi3zVDDpGkrnHCSns1wzIaUzMI/5gEACShezTHvdyhwKGqYVM3CCSfpccyG9GyFf8wDgHF7dPu5N1y697k/MOX8b//r9Xde/1f/u8/2PY4h5NHt5x7770xSc+uPGl6ddde2Oj6Z/I2//l0f/Bt//bs+2Pc4Us0654U50T6O2d1zEusTqpnbq/MrP/W3fjQisl/5qb/1o32Ppc+YpwCM2sV7n7d18d7nZVPOg//6pmzrNfPexzGQbKm58UUNt8paa/uiOj6RfOTOH84+cucP9z6OhLO2eXHRnGgdx+xDZW21SjcXze1WMaefl100TwEYs4v3Pu+m/IfalY/9A5loCgubm9ScpJTjqm11LGPOccwLc0KOIye5PqGauS2rYp4CkIyL9z73Ly7e+7zs2qMvzrI/eqlMLNcefXF28d7nZY9uP/c/RMRf28tfUXMy9hx3batjGWOOc16YE7LO9LE+oZq5LXUxTwFIysW9W2ce/8yPZ9kXXiYTy+Of+fHs4r3Pyz78zh/+7dj9gxHziHiqmpOx57hrWx3LGHOc88KckHWmj/UJ1cxtqYt5CkBS8ltnLt/3D7LsCy+XieXyfbu3xrzunz/j52KxsPkf1ZyMPcdd2+pYxpjjnBfmhKwzfaxPqGZuS13MUwCSs3/rzMWfzLIv/L8ykVy7+JPZxXuflz3y/ud8MxaLmu9XczL2nFRtq2MZU05iXpgTso70uT6hmrkt5ZinACRpcevMC7Ls375CJpLHP/OC3m6LUXNynDmp2lbHMqacxLwwJ2Qd6XN9QjVzW8oxTwFI0uLWmX+YZf/2n8lEcvm+f9jbbTFqTo4zJ1Xb6ljGlJOYF+aErCN9rk+oZm5LOeYpAMnav3XmD2/Jsi++UhLPtT+8pffbYtScHEdOurbVsYwhJzkvzAk5SoawPqGauS3mKQCTsH/rzO+9OMu+eEoSz+O/9+Leb4tRc3IcOenaVscyhpzkvDAn5CgZwvqEaua2mKcATMLi1pl/lGWX/4Uknsv3/aPeb4tRc3IcOenaVscyhpzkvDAn5CgZwvqEaua2mKcATMbi1plTWXb5pyXRXPvDU4O5LUbNyTrTV22rYxly+pgX5oQcJkNan1DN3BbzFIBJWNw68xNZdvlnJNE8/ns/MZjbYtScrDN91bY6liGnj3lhTshhMqT1CdXMbTFPAZiE/VtnPvqPs+xLPyuJ5vJH//FgbotRc7LO9FXb6liGnD7mhTkhh8mQ1idUM7fFPAVgMvZvnfmjn86yL/28JJZrf/TTg7stRs3JOtJ3batjGWL6nBfmhHRJ38dw2jO3pxvzFIBJ2b915vdfnmWPvUoSy+O///LB3Raj5qrzkrvvzF5y951Hfp2d339t9pK778yuPvrqE3vPPtJ3bY+hjs9fOLO/j89fONP7/j9/4Ux26p5zvW+X4x5/vt2rtvlxp895MYY5UU5+vDz9gdvW8npjPqaedPo+htPeGOd2Mac/cFt25kNvPfb3WfUztypDP2aYpwBMyuLWmX+SZY9tSmK5/NF/MrjbYoZac6c/cNv+QrWYk3r/db3f9qfekL3k7juznd9/3Ym9Zx/pu7bb1vGpe84dqKerj/7yiW7z/P12T9b63f9nPvTW2verm39ta7nv8RdzsEF6smPsc14c5tieH6+qcvXRXz727bXz+68rNEiP/nonMafyOd2UocyZodYq3Qx13ZYfk19y953ZqXvOZecvnKmcy6fuObe2Od52frY9/g99HWaeAjA5+7fOfOHns+yx05JIrn3h5wd7W8yQay5v1PQ9jsNm+1Nv3DtB/le9j+W4MpTabqrjq4++JnvJ3XdmZz70tgOPL5o/rzmRbZWP4/yFN/e+38586G2Nc+v8hTcv1W7+2Eltr6OMv+8MYV4c5tieH7O2P/XGpcdSPo4dNuU5ffoDt+1d2Tye7TaEWqWboa3bysfjfF7kc6GYRYP05OfnmGOeAjBJ+7fOPPSKLLvyLyWRPP7QKwZ7W8yQa+709jt2F90tn5dn+3fetPI55+8/u///z3zobVl25V8uX2F48bWV73f14msrrxTK37fu6/tXJxReq+177jz0qweed3r7Hdn277wpO739DrW9oo5Pb79j90St9Pj277zpwP44cNVXaXsXa6ryKrGHfnXp9fOTxnLO33+28/7Px1pXR8X6LtZ2uT6qXqfquflrFD9Xvk3K86vN/KvablWftWqbFfdd2/GXn5fP8arxnN5+x9L2Kz+3uE3L27iqtoYwLw5zbM+3W90+btpXdceicn3sPPSr2al7zu3XVvl4WbU929bHUY6p5ddpUx/52PPHiseafFvmn7Pt8aXq50eb48thfxYMoVbpZmjrtqpjbD5vmuZu1ZwqHy/qjsldji3595fHkD9v1TGjOF/LPwuq5mZ5XhbHWffz0TwFgAqLW2d+LMuuvFoSyeWP/thgb4sZcs0tTsrrn7M4sd3973whu7tY330sX/yWn7NY3B58zXwBfPXi65a+dvXi6woL7oOP7S58F889eILc/Fmb3jN/neJ75p+h+NmnXNt1dVy1v5pSbKzl33P+/rON27lq3+W1mz+Wj6NuLE37Pz+pK35f/lixtvL3LD7v1D3nDsyF8vev+kxVr198rM38q3osf/3inKka05kPva1y268afzHl9y5/rfiZquZy1bZu2pdDmReHObYfbJDWH8vyY2pxm5SPs3WP5XVUfo/8a3VzrUt9rJpTVfVXnrNt6qN8fCmOv+r4v+r4kn99t7F08HvK87rN9m+TIdQq3Qxt3ZbXXpv1yKl7zh2o71U5+A9z3Y4txflZbGxWHRNWHTOK68by3Cz/nC9+vuL7HmUbm6cwYLe86eMPnjr7wMMiUp9b3vTxBw87x/ZvnfniL2bZH/+KjDzXvviLg78tZqg1t98grfn6mQ+/vfLr5+9/y97J4xuy7U+f2f//dc+p+/6rF1+/9LWdhw7+nr4zH3577fjq3rsqTe+5v+Cu2D5Vj0+1tqvq+OrF1++dIL2l1WfK91nb5xdrYvvTZw68Rv7fbV67bv/n46+qs1P33HFg/1fNl92GzR2t5055POUUX6vN/KsaZ914y1cDVc2FtuMvz5+6eVo1t4rPr6uffJ9X7cshzYuux/ZVtbv96TO1zylvq7pjbL7tqo6LdfXatT7azKmqfVeuh1X10WX8q44B+VxZNV/bbv82GVKt0s3Q1m1VPy+q5mjdz4NVr12cc22PLfmcqPr51eWYUXzt4vwq/+yv+7m0ai1rnsLIvfzWC9mFRx4XkYa8/NYL2WHn2P6tM5/9qSz749fKyPP4Z39q8LfFDLXmTm/fvreorH7s9Pbtewveg9+389De79H79Juz8/e/dW/B+6sHnrP96b3fsfjQG5e+v+57yl9fPhlo/x5t3/PqxV/dW5S/tff9MfTarqrjqu2X10cxXfZZ3f7f/vSbD70v676naqzLJ3zt5ksxZz58W+Xj5fEUt0M+/vw928y/7I9fu3dye1urz1z1Gau+d9X4i6l7jfxrp7dvr32s/Fna7MshzYuux/a8/suftzgv6uq/vL+67KNivVbVVNf6aDOnyp+x6v1X1UfX8a86vjRt1/yztN3+bTKkWqWboa7bsj9+bXbqnjuWfrYWv1Y3f4rzs6mu2x5b8uN0nlXrqKa1Xz6uqrVE/ljd/D/MsdA8hRF5+a0Xsi989S9FpCFHbJDu3Trz/Cy7+joZeS5/9PmDvy1mqDW339ypeWx/0Vn6vu1P790e/Nk37f//7U+fPfCc/Ht3Pvumpe8///G9RfKlW1uNM3/+6e3ba8fR9jWq3nP/5KD0+NVLt1Z+/qnWdl0d75+Qtdjmq/ZZXjfFetr57JsOPFb3Gvnzzn/8ra33f76Pq76n63yp+hyr6rH8GYrjbDP/mrb/qjHktV81jjbfu2r+5F8rj634WL79y9+ff8aq/TKkedH12F53vMz3e5s5sqqm8/1Xte32GwwtxtpUH4edU+VaXVUfXce/atvVzZUur9ElQ6pVuhnKui3/uVaeZ3Xzr6nG89ouH2/Lj7U9thTnev6zsO5nwarXrfr5XX4s/7lU9ZmPslYzT2HgNEhFVmevQfp9EfG9h5ln+7fOXP7lLLv6ehlprl3+5dHcFjPEmls0d6ofu3rpjYWT1dfXPrZYnO7+985nzxROrM8sve/5j79tb5H7xtqvlR9fLLxfv/Q+259+y4Hx755Ad3/P4utUfS61XV3H259+y95JzNuW9ll5m+fPraqLuv2Xv05x/+zfKluxv8rjWLX/F03ZxevndV58j1XzZdX7FWs4/3p5Oywak+3nX9X2r3qsatvUbZdV4181N4tfK4616rHFP6acOfCcqvEOcV50Obbn+6VYa/lxrPhZ81qumjvFbV2u0ab5WDe/DlMfbedUcZ/W7edV9dF2/MXPXnd8qdqGVdu77fZvyhBrlW6GsG7L6648J6p+BlbNkeI6qWrOHmxsHpyHq44t+c+j/L/zOV43R5uOGVXHwfJjVT//8tc87FrNPIURaGqQPvfmLHva07LsF1797QOPf+Khv8ye9rRsP0dtPhVfqypdXuvu7W+1+r5fePW3s6c9Lcue+YOrX7/8eZ/2tCx77s1H/9wnkWf+4O54X/aKbzd+vbyP227buhqpq6W6cdTl7LlvHdhHbffvurPXIJ3vpbPFrTP/PMuu3iojzeOf/eejuS1mSDV3evvgXxStSvH5xdu5dherb2/9mjuffXOWXb01O//xtze+36rnVb1n3fObvlb1ntnVW7PtTy/fVrm7yFfbRU11XLWNr156U+32zXN6+1xjve189s2V+6Rcc+X3OOr+L46//F5XL72p8rGmOZF/zlXzr/w6bebf1UvLf32++Pnq9k95rrcZf9NrFZ9XNZfq5lfbeT/EedH22N40B8r7Krt6a3bmw8t/bbrqmFSuj3wutD3ed6mPo86punpcVR9Vn/Gwx5equVLeZl22/9iO4bQ3hHVb/vOv+HNwVS02HUfLdV1+3eLPn6ZjS/FYUHz9qp9Xq44Z5TGc+fA7Kh877M8y8xRG7jAN0nU2R8uvN7QGaf68unzioZNvenZJcfxN26rN5zjpBmnVPhpxg3T31pmPbWTZn7zxSDnz4dsPLh4efFvj8/d/kB/ivXYeXvy1xtPb51p9z+ntc9mZD99+6M9XPLla9dlOOpc/tjGa22LWWXNjSL4ovvqHb+59LGPMUGt7anUs/ST/WVf+mTPEeWFOSFWGWKt0Y26PI6e3z+02jM1TSFPXBml+xeE6m4P56929/a0jv9Y6G6TF1ypug+LjQ7+StHj1a3n7drmKtsu2baqlozZI+8pRG6QRhVtnvvS6LPuTNx05Ow+/pXBCV/+8RYP08O+1aJCufu6pe97Z+rl12X7wba0+20nm2pdeN7rbYtZdc0PO6e1zR667qWbotT2lOpZ+cv7jty39nBzyvDAnpJgh1yrdmNvDz+6VqreZp5CqLg3S42iOfuGr3RqkZ88tmnRVDcq6Jt7LXrG4kvKZP5i1ar7l31PV1PuFV3/7wHuXm6nl7VR1m35TA7JuO7d5Tjl1t9lX3V6f7/PitmratnVXkBZr5WWv+HZlg7Tqs5w9960DzdFifuHV325s0hbfs6qeio8Xn9umyb2eBunerTMP/3SW/embj5ydh9+610R8+1peryn7DdIWz91vkB7h/bYffPuJfba2efzhnx7dbTHrrrkh5dQ971y+rXEA4xpjhl7bKdex9JMzH7l9+XbJ0nOGPC/MCSlmyLVKN+b2sHL1j95S8SsEbjNPIWVtG6TFxtk6rvQ8TIO02OSsazZWNdCavq+pQZo30dp83qpmX/7a5aZuVROvqoFa/hxtnlOVVbeq5w3Wuu2UNzXbNkjLjcqq12r6LJ94qFuDtOm1iuOqe05Vg/eYGqR7t878eJb96VuOnJ2H8ybibZVfLzeQrv7R2yqfd3q7/Pt7Fr8v6MxHzmXZn+a/sP2OA889//F3HHidMx+p/71j5ecebGwdfN3tB2/Lth+8reZ3e92RZX/6ltqvt/m8h83lj/346G6LWXfNSZoZem2rY+kjQ54X5oQUM+RapRtzO92YpzASbRqkXZqKR2mQViW/orDYGMsfK35vUxOvqVnW9Fm6NITrbsevGmNx25abqMXx5I3G/PO2ec6qseWfJW9AFq+eLG+3cmO1TYO06r2Kj+XbIH/tqn1Svoq07VWsxedVXcVbtc/z7111Fek6GqQRhVtnHrs1y/70rUfKokH6jsbnnf/47XsNw7cvfS1vopZf8yV335ntPLx4fvmxq39U/967V5DeUTmW/PuKX8/HVxzH9oPvqBxHcbz5WHebuG898Bp50/Wo2zjPtcduHe1tMeusOUkvY6ltdSwnmTHMC3NCBlirGxFxYUU2Jvw6rZjb6WVg8xRo0qVBWmw6NTXkjqNBWtXQq2oalhtodbdkt7nF/rAN0uIt73XvX7zq8RMPLV+BWtWwa/OcupRvp2/66/XlKzK7NEjr9lPT7yAtX7natUFat5/KzeOq2q0b7/E1SPdunfmDn82yP3v7kbLzB7tNxO3fvb3xeec/sdcg/cJtBx7f/t3dx3f+4B2Vzy8+/pK778xO33vHgeftNifvWHq/U/e8c+m5ec58ZPdK0fLjp+89+Hg+tvOfOPjZrn7htv3H88+fj7P4/NP33pGduuedR97GeR7/g58d7W0x66w5SS9jqW11LCeZMcwLc0IGWKubEZGtyOaEX6cVczu9DGyeAk3aNkjz5lOxmdW18VnXVG3TiKz7faCrGqJ1DcqqqzHrxtWmGXyY96m7qrOc4nZp85yqFJuN5eZs1f5ed4O0vP+qfiXBYRqkdZ+lamxN23PVVdFrbJDu3TrzT7Psz95xpOz8we17DdJzjc87/4lzew3S21s9vv275/Yaj4vHdxuk7zzwvKrHsj97x16DdPnx7M/ekZ2+9517jcuDjy8apwfHUPXZ8sZs/vmLj+djrnufw+byx/7paG+LWWfNSXoZS22rYznJjGFemBMywFrdjGE1JIf2Oq2Y2+llYPMUaNL1r9gXG1Kr/iL5OhukfVxB2nTV49lz31p5ZWPT402NvXIjtO7qxjbPqRpH3qwsfk/x96SWG7bHcQVpfnVn1S3+x30FaZ8N0ojCrTNX3pxlX7790Nl5pNBEbHje+U/csdcIPfi8A03IwuNnPrJ7G/vOI4vH95uhhedVPZZ9+fZFg7RiLPlrl8eyf+v8irHl73vmI+/Mdh45d+B9it+/3yA9wvbNc+3Km0d/W8y6ak7SythqWx3LSWRM88KcmHYGWKuzvTH8/Yh4YU3+bkTcEBFPnuDrtGZup5MBzlOgSdcGafFqwDZXLrZJm9fq43eQ1v1e0eLjeYOv6a+rVzWUy783s6ppWW7etXlOU8p/PKnqVvOq3895HL+DtDyGqv1bdfXtOn4Haf8N0vzWmZ/Psi/fcejsPHLHXhOx+XmLBuny107ds/d7O0uvudsgXTxv0Qxtfiz78h17zcnl19z+3Tuyq19Y/r58fMVxbP/u4rHiuE/f+879x3Ye2f3e4tfqxnCUPP4HPz/622LWVXOSVsZW2+pYTiJjmhfmxLQz4FqdxWI8dZlN+HVWMrfTyYDnKVDlMA3Stg3GdTZIy43ZcpoaaHW3pbcZf9P3FptvTQ3So/4V+3z7t/1r7W0/S93vSq3bRm3/in3drfrFBmnTc8oN0uJnXMdfsR9Ag3Tv1pkXZNlX7uyUnUcO/mX6pb8W/8A7s+wru//b9LydR965/5qn721+TvGxU/fcWftYMXXjqvv6foO0NJ4zH6l+r+2ddy699m7jdfk1um7jci5/7AWjvy3mKDUnq7OuWjvpjK221bGcRMY0L8yJaWfAtTqLYTUkh/Y6Kx3H3C6unfM1dt81XE7b9VTVucvVf3u415rwPAWqHLZBus4/2NS2QVrVNCvfyl3XqCw2V5/5g1mnKy+rGnBt37fpNZq2RVPj8zDN0fIYq27JL+7TfLsUm6ltG6Rf+OrBq1Vf9opvV/66gnJTNH//qitt88ebtnP5CtlyPQ2lQRpRuHXmj9+eZV+5a1A5/0C+0Di59zx9b9787P/zF3Ptj9+ezG0xR625nUe2ahvbu43p/vdXX1kswvsfS+q1fdzHzvz4V/0PUOPax9I9Y5wXXeaE+k4nA6/VJ0TE9SvyhAm/Titdf96Vm4b5um3nka2l5y4apP3Xcnlcq9ZT+cUT63itic9ToEpTg1REdrP+BunerTOPbGbZV84PKqfvzRteJ/eei5Oz/j9/MY8/spnMbTHrqrlFQ7T5MRl2xlrbJ3HszE8qi8ekqsckvYxxXnSdE+o7jYyxVummy9xeNAMXj+W/Rmt7Z6vy+Wc+svz4GHLqnvyiiv7HYp5CgjRIRVbnGBqku7fOXHhhln3113pNvoBaXA24dazvd/WL76q4cuV43/OwuXzhhcncFrOumquqkTP37Z5cVz3WVFc7n7trqQ4O3C71xXcdeOz0vVsH/vvUPQffs1zLdXVVft5L7r4zO3Pfwedu79y19Jx8THWvU/xa1Xare+7pe7cqt8FxzsWx1vZJHDvzuizXz6l7lvdJm5prU+flGij+d7E2q46fO5+7a+k9q563e6J88LnleVo1r8o1WvU6bcc/9IxxXnSdE+o7jfoeY63STdu5nc+58hokn5tVc6ipdqvm1nHPwzbrqarXqTo2tV2bVa3zmrZF+VhX/pzmKSRCg1RkddbdII0o3Dpz9VyWffU9MrBcu3ouudti1lFzi8bd7n9v75xfeixfnO587l1Lj5W/7/wDd+0/lp90Fl+r/P3F19hdoJ7Prn7x3Utfyx8rv9aiCXDwsTP33bU0tuL488eufvHdS2M7/8BdtV/b+dy79k5E7lp6/vbO+aXPXtwe5XGp7V3HfezM91lxX5y5764Dj7WtuS51XnzN/Gv5WK5+8d2Vc63q9fPXqXqsWHP5ZyqO4cx9d+2duB6c88XHqmq6zfiPY1+ZF7u6zAn1Pf76HnOt0k2buV21rlmVqhqvS9UaZ93zsOm9qj5v3fe3fa3yMa/4WHHtVzwOFZ9//oH6z2CewshpkIqszvE0SPdunfncL2XZ135dBpbHP/dLyd0Ws46aq/6X+7v2v779md0Txu3PvOvA9129/J4Dzz11z+5VOFWvX/X4qXv2rlKtGFO+qL16+T0HHs/HsvO5d1WO/9Q9y++Tfe3X9xfV+1cJlD5L3fPL79/0OfMmQt1/54/VjXHKtX3cx8686bF8Bc2iDtrWXJc6z+dI1fPz16qqh/0T0Jrxn7nvrsrXK181V1W/5dcu13xxm6wa/9Az5nnRZU6o7/HX95hrlW7azO26ediUprlTd8woroXWOQ+r5l7T8+qOFW1fK5/LVWOu2pb5ca+41jVPIWEvv/VCdnrrM3++ecen/5OIVOeYGqR7t868KMu+9l4ZWC5feFFyt8Wso+Z2TxDvyrKvvTe7ennRcMy/fv6B8wcWw+Wcue/8/uvk/7+Y3cXpXa0fz7723uz0vdW3wy8anL+2/9ydz7175XPqXnN3Qb78/vlnvnr51yu3V9XnLH9P/n5VY1DbBx33sTOvkfMP7O637c/kv1piUX9ta65LnefzKX/fqlpqSrH+6uZhuUab5mj+OavqPt9GxXmzavxDz5jnRZc5ob7HX99jrlW6aTO3m9ZHdambu03zq7xOWtc8rHrNqvVU8fPWrcfavFbdOrBurZcfI3c+927zFKbg5W+4MH/5Gy7M//bf/39uEZH6xJobpBGFW2f+5K4s+9rdMpBc+5O7kr0t5qg1tziR3v3vnc+9p7CgvDvb/sy79xaS72l8ncUCd/HY1cvvXXr94vOrHs++dnd25r7ze43E7p8nH/+q788bBucfeNfS184/sHeb5eX3th53ecyLZujye6rtZcd57Mxroriv8328e1LXvua61Hn+eFWNrZoDq5KPf9X353Mhn791n7Nqnq8a/5CTwrxoOyfU97jrO4VapZtVc7tpvp66Zyvb/sy7K+dCPt+rXqv4Pfkxo+p11jEPq16jaj3VdNxpGk/5tbrO5bbrWvMU0jMXkVZZm/1bZz7/y1n29ffJQPL453852dtijlpz+4vfwmPnP7l3+9En35VlX39fduq39v5gwJf+9f5z8gXmmfvedeC/8+/Jvv6+/YZg+fXz16x6PPv6+7KrX/rXu03a39o68PiZ+961WNR//X3Zzud3rwDa+fyvH3je6XvvOvC9p++9a+m98u/NX6vq8xc/b/lzFz9n02cvj6v8mNredZzHznxfF/dPcX/sfP7XW9dclzrPX7P8vnVzqDg3ivWX12PV3C1+b9V7lWs5H1NxrFWPtRn/kJPCvGg7J2PPAz8AABO3SURBVNT3uOs7hVqlmzZze/8feivma3nNUzVfivO3PO8XV5Au1j/rnIdtv1Y8JpTHWJWm1yofy4pzvPza+w3Siu1onkLa5rH7LxsiUp95rNHi1pkXZ9nXf1MGkssXXpzsbTGHrbmdz//G0m1I2595z/7X85PY3YXlb+4vPg/ctrT3tabXXJyYHnzdqux8/jcOvF7Vc3cX6u3fL/v6b9beXlp8v/OfrP5Lr1XPvfqluxu/Xn6/q1+6u/Ixtb1wXMfOqv2ab/vifszralXNtam7qvoopvhadc8tzsW62jz/yV878Fpt3q9ubhVfq8v4h5oU5kWbOaG+x1/fKdQq3bT9eVe17iquHZrquGndVZzj+TpunfOw7XqqzXqwy9ps+zPvqXxOvs3qvl7eXuYppOt6EWmVtdq/debP3p1l/+4e6TnX/uzdyd8WM+Sae8ndd2ZnPvprvY8jxaRW20Ou41VR58NJSvNiKHNCfatV1msoc1vMUwA4VotbZ16bZf/u30jPefzzr03+tpih1tzOo7tX6uw8enfvY0kxqdX2UOtYnY8rKc2LIcwJ9a1WWb8hzG0xTwHg2C1unXlJlv3790vPuXzhJcnfFjOUmtv+veW/brrz6N2910CqSa22h1LH6nzcSWle9DEn1Lda5fiN5eedmKcAcGT7t858+b1Z9u+3padc+/J7J3NbjJqbVlKtbXUsR0mK88KcSDMp1irdmNvDj3kKAGuwf+vMo6/Psn//Qekpjz/6+sncFqPmppVUa1sdy1GS4rwwJ9JMirVKN+b28GOeAsAaLG6deWmWPf7b0lMuX3jpZG6LUXPTSqq1rY7lKElxXpgTaSbFWqUbc3v4MU8BYE0Wt878ZpY9/mE54Vz78m9O7rYYNTeNpF7b6lgOk5TnhTmRVlKuVboxt4cb8xQA1mj/1pmLb8iyb9wnJ5zHL75hcrfFqLlpJPXaVsdymKQ8L8yJtJJyrdKNuT3cmKcAsEb7t87c/9Is+8ZH5YRz+f7p3Raj5qaR1GtbHcthkvK8MCfSSsq1Sjfm9nBjngLAmu3fOvOVf5Nl37ggJ5RrX/k3k70tRs2lnanUtjqWLpnCvDAn0sgUapVuzO3hxTwFgGOwuHXmTJb9+cflhPL4xTOTvS1GzaWdqdS2OpYumcK8MCfSyBRqlW7M7eHFPAWAY7C4deZlWfbnD8gJ5fL9L5vsbTFqLu1MpbbVsXTJFOaFOZFGplCrdGNuDy/mKQAck/1bZ772wSz7i0/KMefa1z44+dti1FyamVptq2NpkynNC3Ni3JlSrdKNuT2cmKcAcIz2b525dDbL/uJ35Jjz+KWzk78tRs2lmanVtjqWNpnSvDAnxp0p1SrdmNvDiXkKAMdocevMT2bZf3hQjjmX7//Jyd8Wo+bSzNRqWx1Lm0xpXpgT486UapVuzO3hxDwFgGOW3zrzHx97d/af/+T/k2PKf3zs3W6L2aPm0spUa1sdS1OmOC/MiXFmirVKN+Z2/zFPAeAE5LfOjCnv+lf/Z7b1mnm29Zp59sj7n9P7eLrEbTHjrDk1rLbL1HH6Nb2OTGleXEx4TkyhvqdUq3ST8twe2zw3TwHgGF364HPmD//W//2pT/3G33t4LPmuJ1z3rYjIIiLbvv1Zn+t7PF3ysz9xwy0x8dtixlhzalhtl6nj9Gt6HZnSvEh5TkyhvqdUq3ST8twe2zw3TwHgZHx/LH7gDj3fjL0FTEQ8ewDjOUzcFjOumlt3Uqjhukyttqdcx8WkXNPryJTmRYpzYkr1PaVapZsU53YxY5rn5ikAHKOnRv8/7NumuID50Rj+IqYqbosZV82tOynUcF2mVttTruNiUq7pdWRK8yLFOTGl+p5SrdJNinO7mDHNc/MUAI7R9RFxw0jyX2OxgHlpRLwwIp42gHF1yfXtdkvSxlRzalht15lyHU+lpteR62M6ro/+t7f6VqusX4pze6zz3DwFACJisXjJYvEvqdf1OB7oSg2TGjVNytQ3pM88BwBGxwKGsVPDpEZNkzL1DekzzwGA0alawHxHj+OBrtQwqVHTpEx9Q/rMcwBgdKoWMDAmapjUqGlSpr4hfeY5ADA6FjCMnRomNWqalKlvSJ95DgCMyhNisXi5FhYwjI8aJjVqmpSpb0ifeQ4AjM71sVjAfDMsYBgfNUxq1DQpU9+QPvMcABgdCxjGTg2TGjVNytQ3pM88BwBGp2oB83f6HBB0pIZJjZomZeob0meeAwCjU7WA+f4+BwQdqWFSo6ZJmfqG9JnnAMDoWMAwdmqY1KhpUqa+IX3mOQAwOk+OxQLmG2EBw/ioYVKjpkmZ+ob0mecAwOjMYrGA+WpYwDA+s1DDpGUWapp0zUJ9Q+pmYZ4DACMzCwsYxm0Wapi0zEJNk65ZqG9I3SzMcwBgZGaxvIC5scfxQFezUMOkZRZqmnTNQn1D6mZhngMAIzOL5QXMM3ocD3Q1CzVMWmahpknXLNQ3pG4W5jkAMDKzsIBh3GahhknLLNQ06ZqF+obUzcI8BwBGZhYWMIzbLNQwaZmFmiZds1DfkLpZmOcAwMg8NRYLmKthAcP4qGFSo6ZJmfqG9JnnAMDo3BCLBcwXwwKG8VHDpEZNkzL1DekzzwGA0bGAYezUMKlR06RMfUP6zHMAYHSqFjBP73NA0JEaJjVqmpSpb0ifeQ4AjE7VAub7+hwQdKSGSY2aJmXqG9JnngMAo2MBw9ipYVKjpkmZ+ob0mecAwOg8IxYLmIthAcP4qGFSo6ZJmfqG9JnnAMDozGOxgHk4LGAYn3moYdIyDzVNuuahviF18zDPAYCRmYcFDOM2DzVMWuahpknXPNQ3pG4e5jkAMDLzWF7AfG+P44Gu5qGGScs81DTpmof6htTNwzwHAEZmHssLmBt6HA90NQ81TFrmoaZJ1zzUN6RuHuY5ADAy87CAYdzmoYZJyzzUNOmah/qG1M3DPAcABuqmiHhVRFxfevyHYrGA2YnlBcwsIjaPfXSwmhomNWqalKlvSJ95DgCM0pWI+POIeEXhsZtisYD5ZCwWMLOIuHPv8Y2TGyI0uhJqmLRcCTVNuq6E+obUXQnzHAAYmY1YLFYei4jnx/IC5uaI+K3CY1d6GCfU2Qg1TFo2Qk2Tro1Q35C6jTDPAYARuhKLxUkWEV+r+f95NvoYJDS4EmqYtFwJNU26roT6htRdCfMcABiZjVhepNTlSi8jhGYboYZJy0aoadK1EeobUrcR5jkAMEJ/Ee0WMDf1NUBYQQ2TGjVNytQ3pM88BwBGZzNWL17u72ls0MZmqGHSshlqmnRthvqG1G2GeQ4AjMz1sfpfeed9DQ5aUMOkRk2TMvUN6TPPAYBR2oz6xcuD/Q0LWtsMNUxaNkNNk67NUN+Qus0wzwGAkWn6V955f8OC1tQwqVHTpEx9Q/rMcwBglDbD7wZi3DZDDZOWzVDTpGsz1DekbjPMcwBgZKr+lXfe54CgIzVMatQ0KVPfkD7zHAAYpc3wr7uM22aoYdKyGWqadG2G+obUbYZ5DgCMzMsi4r/F7gLmsxFxQWRkeSDUsKQVNS0pR32LpJ+meb4RAAADtBn1f21SRERERERkXdkMAIAB2oz+F0oiIiIiIpJ+NgMAYIA2o/+FkoiIiIiIpJ+3R8QNEfHkAAAYkM3of6EkIiIiIiLpZyt2/6r9LAAABmQWu4uUmyPiFhERkQGm6iT7Z/bS99hERKR9bg4NUgBgwGaxu1gREREZWqoapPnJdt9jExGR7pkFAMAAzaL/hZKIiEhVNEhFRNLKLAAABugJEXG9iIjIAFPVIH32Xvoem4iIdM8TAgAAAGitqkE63wsAAAAAQNI0SAEAAACAydIgBQAAAAAmS4MUAAAAAJgsDVIAAAAAYLI0SAEAAACAydIgBQAAAAAmS4MUAAAAAJgsDVIAAAAAYLI0SAEAAACAydIgBQAAAAAmS4MUAAAAAJgsDVIAAAAAYLI0SAEAAACAydIgBQAAAAAmS4MUAAAAAJgsDVIAAAAAYLI0SAEAAACAydIgBQAAAAAmS4MUAAAAAJgsDVIAAAAAIEmzFs85TIO0zesCAAAAAPRqFhFXIuL5Dc/p0iDdiIjHQoMUAAAAABiJrdhtej4W1Y3SNg3Sjb3vz/ZeDwAAAABgFGZxsPlZbpQ2NUg3YtEYzTM7gTEDAAAAAKzNViw3QfNGaVWD9DUR8ZWKx7dOdtgAAAAAAEc3i+pGaNfMTnbYAAAAAADrsRVHa45unfSAAQAAAADWZRauHgUAAAAAJmwrXD0KAAAAAEzULFw9CgAAAABM2Fa4ehQAAAAAmKhZuHoUAAAAAJiwrXD1KAAAAAAwUbNw9SgAAAAAMGFb4epRAAAAAGCiZuHqUQAAAABgwrbC1aMAAAAAwETNwtWjAAAAAMCEbYWrRwEAAACAiZqFq0cBAAAAgAnbClePAgAAAAATNQtXjwIAAADQp1ve9PEHT5194GGRqtzypo8/eMwlOF/ni02tnk9g/wAAAACk7eW3XsguPPK4SGVefuuFrO8a7WJq9Ty2/QMAAAAwOC+/9UL2ha/+pUhlxtaAm1o9j23/AAAAAAzO1BpK0i17Dbjvi4jv7btW25haPY9t/wAAAAAMztQaStItew24eaz5d4Uel6nV89j2DwAAAMDgTK2hJN0ytgbc1Op5bPsHAAAAYHCm1lCSbhlbA25q9Ty2/QMAAAAwOFNrKEm3jK0BN7V6Htv+AQAAABicqTWUpFvG1oCbWj2Pbf8AAAAADE7XhtJzb86ypz1tkV949bcP1djJX+ew39/36x9XfuHV3z6wfcvJP0/58929/a3954ygATePiB+oyZEctUH6slcsb/+mGjp77ltLz3/uzcv7oO1+Hcj+AQAAAJiOLg2lcnP0KM0dDdLqTKRBmjXkSI7SIG3a7s/8weXtWjcf8nzioe77dSD7BwAAAGA62jaUPvHQXy41fopNn6k0MI87+TatugqxKWNskP7AD/zAfqLnBmmx2VlsbBa368tesajVYu2fPfetA69V1VQ97H7taf8AAAAATEfbhlKxUZQ3kJqapuVGULmZVGyQFptTxe8rvn75VuZPPHTw6+XGVlUDtvgZqr6n7XO+8NXlW7HLjd7i+5efu44GadsrSMvbrdjk67tBWhSLBulf28uhHKZBWtx2d29/q/HreS2suvLz/2/vjnHktgEogLJ06TKljuADJNg5Qyp34wNsAN9gFjlA3KRIkQB7k9kuRYqkzgnSuXCbYFJkmRAESYkjyaOs3gN+M+BIsyLdfFNSvt4VpAAAAAAb1VMoTSnaegvSUuJ38wI0v+W5dRt0XiC2jlUqZGtjfv/jz+K58795yt+3dkFaej5mb0l3g4J01rmuKUjjNS/dRp/P+fc//lUsTPPEUjz+O1GQAgAAAGxUT6FUKtzyUumagrR0/HyHaCya0nIqnqNUWOUFYjx2+nvT0mvqmPTW6niu9Pylv6/nkQStZ1W2CuBSQRp/e7orsvSZgnS8vEyv95THGeTHnDqvClIAAACAz+yaQqm1w/KaW+zTcWkhmRaksdCb+lmrQKyVYVPG1H53vmMwjkt32vYUa3ML0tZu2NLvV5AqSAEAAAB2ac5bv9Mdn7Fw22pB2iqp0t2UY2PyHaW1QmzsGai1W7OXusW+9CxVBWn7mrvFHgAAAGCHphZKeeHTWwRtoSAtlWKt0qo0pncH6RYK0hsUcO9CCOdGWgXpr88pfe/dUus5zdhLmvJHP6TrubS+8peLKUgBAAAANmxqoZSWRGmJVNtBmhZz6WetZ3S2nkE6tyCNBWZaUOU7B6eM6X0G6a0K0tLclK7TSgXcQ/iv8KymUpC28rDUeq5dz/zapNc1/c+BdB3kJWn8PN2RqiAFAAAA2KieQqn29va0lBu7tXvKW+xjEbVkQdp6JmfPmNZ1KL3F/pYFae1xAZ/hLfYPIYTL8Xi8nM/nalKtccfjcfWCNC02Syndft9aw/kcK0gBAAAANqq3UCqVQvmY/G33Yzss02OmBdIat9i3is+eMXG3aW3MFgrS0lzkj0hYsyA9nU6XJZxOp1iQ/hBCeBNC+GKp9Vy7/mPzX5rTVgGtIAUAAADYqLmFkrzsbKwgfXz+HYP1PGt+AAAAAIj2VihJX+YUpMMwXA6HQzWpu7u76rhhGBSky84PAAAAANHeCiXpy8Ze0vQYFKRLzA8AAAAA0d4KJenLlQXc8Dz+bQjhfSGtgrQ0PuZtUJAuMT8AAAAARHsrlKQvMwu4IflumlZBWhqfZ7CeF5kfAAAAAPZWKElfFKTbjoIUAAAAYKa9FUrSl5kF3KsQwutCphSkpe/FvLKeF5kfAAAAAPZWKElfVirgphSkV9nbelaQAgAAAMy0t0JJ+rJmQfr09PRvgoJ0S/MDAAAAsB/3350v3z7+8vHhp58/ieRZsyCtZNa59raeFaQAAAAAM91/OB/uP5wPX339zXuRUsLyBdzTc34rZNa59rie514zAAAAAP5xEBnJ0tY8V+vYLzUAAAAAzHAIIXwpUskhLO/QyBLHvvU1+7/PDwAAAMCuvBYZydLeNDLXra/VS5gfAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAgM35Gw6WmGQap8I8AAAAAElFTkSuQmCC"}},"execution_count":null},{"metadata":{},"cell_type":"markdown","source":"## Import Libraries","execution_count":null},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import re,os\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\n%matplotlib inline\nfrom colorama import Fore, Back, Style\n\nimport lightgbm as lgb # CLF1\nfrom sklearn.linear_model import LogisticRegression # CLF2\nfrom xgboost import XGBRegressor # CLF3\nfrom sklearn.naive_bayes import GaussianNB  # CLF4\nfrom sklearn.ensemble import RandomForestClassifier # CLF5\nfrom sklearn.linear_model import LinearRegression # CLF6\nfrom sklearn.linear_model import Lasso # CLF7\nfrom sklearn.linear_model import ElasticNet # CLF8\nfrom sklearn.neighbors import KNeighborsRegressor # CLF9\nfrom sklearn.tree import DecisionTreeRegressor # CLF10\nfrom sklearn.ensemble import GradientBoostingRegressor # CLF11\nfrom sklearn.discriminant_analysis import QuadraticDiscriminantAnalysis #CLF12\nfrom mlxtend.classifier import StackingClassifier # SCF\n\nfrom sklearn.model_selection import GridSearchCV\nfrom sklearn.model_selection import KFold\nfrom sklearn.pipeline import Pipeline\n\nfrom sklearn.metrics import roc_auc_score,roc_curve\n\nimport warnings\nwarnings.filterwarnings(action='ignore', category=DeprecationWarning, module='sklearn')\nwarnings.simplefilter('ignore')\n\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.model_selection import cross_val_score\n\ndef seed_everything(SEED):\n    np.random.seed(SEED)\n    os.environ['PYTHONHASHSEED'] = str(SEED)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"FOLDS = 3\nSEED = 123\nSetup_Parameters = True\nseed_everything(SEED)\nfile_add_list = [1,2,3,4,5]\npesudo_label = True\ntest_pipeline = True","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Import Metadata","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"BASE_PATH = '../input/siim-isic-melanoma-classification'\ntrain_metadata = pd.read_csv(os.path.join(BASE_PATH, 'train.csv'))\ntest_metadata = pd.read_csv(os.path.join(BASE_PATH, 'test.csv'))\nsample_submission = pd.read_csv(os.path.join(BASE_PATH, 'sample_submission.csv'))\ntfrecord_number_df =  pd.read_csv('../input/stacking-data/Image_Name_TFRecord_number.csv')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Metadata-Size","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"print('Train data shape : ',train_metadata.shape)\nprint('Test data shape : ',test_metadata.shape)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Metadata-values [train_metadata]","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"print('Unique values in column with frequency : ')\n\nprint('\\nsex : ', dict(train_metadata.sex.value_counts()))\nprint('\\nage_approx : ', dict(train_metadata.age_approx.value_counts()))\nprint('\\nanatom_site_general_challenge : ', dict(train_metadata.anatom_site_general_challenge.value_counts()))\nprint('\\ndiagnosis : ', dict(train_metadata.diagnosis.value_counts()))\nprint('\\nbenign_malignant : ', dict(train_metadata.benign_malignant.value_counts()))\nprint('\\ntarget : ', dict(train_metadata.target.value_counts()))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Metadata-values [test_metadata]","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"print('Unique values in column with frequency : ')\n\nprint('\\nsex : ', dict(test_metadata.sex.value_counts()))\nprint('\\nage_approx : ', dict(test_metadata.age_approx.value_counts()))\nprint('\\nanatom_site_general_challenge : ', dict(test_metadata.anatom_site_general_challenge.value_counts()))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## One-Hot encode Training Data","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"train = train_metadata.copy()\ntrain['age_approx'] = train['age_approx'].fillna(train.age_approx.mean())\nsex_code = pd.get_dummies(train.sex, prefix='sex')\nanatom_site_general_challenge_code = pd.get_dummies(train.anatom_site_general_challenge, prefix='anatom_site')\nage_aprox_normalized = (train.age_approx-train.age_approx.mean())/train.age_approx.std()\ntrain_coded = pd.concat([train.image_name, sex_code, age_aprox_normalized, anatom_site_general_challenge_code , train.target], axis=1)\nprint('Shape : ',train_coded.shape)\ntrain_coded.tail()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Add OOF Prediction Value to Train Metadata","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"def add_OOF_pred(train_coded,num):\n    for n in file_add_list:\n        df_ = pd.read_csv(f'../input/95-cv-oof-submission/oof_{n}.csv')\n        train_coded = pd.merge(train_coded, df_[['image_name','pred']], on=\"image_name\",how='right')\n        train_coded.rename({'pred': f'pred_{n}'}, axis=1, inplace=True)\n    return train_coded\ntrain_coded = pd.merge(tfrecord_number_df, train_coded, on=\"image_name\",how='left')\ntrain_coded = add_OOF_pred(train_coded,5)\ntrain_coded.to_csv('train_coded.csv',index=False)\ntrain_coded.tail()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## One-Hot encode Testing Data","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"test = test_metadata.copy()\ntest['age_approx'] = test['age_approx'].fillna(test.age_approx.mean())\nsex_code = pd.get_dummies(test.sex, prefix='sex')\nanatom_site_general_challenge_code = pd.get_dummies(test.anatom_site_general_challenge, prefix='anatom_site')\nage_aprox_normalized = (test.age_approx-test.age_approx.mean())/test.age_approx.std()\ntest_coded = pd.concat([test.image_name, sex_code, age_aprox_normalized , anatom_site_general_challenge_code], axis=1)\nprint('Shape : ',test_coded.shape)\ntest_coded.tail()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Add Submission Target Value to Test Metadata","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"def add_submission_pred(test_coded,num):\n    for n in file_add_list:\n        df_ = pd.read_csv(f'../input/95-cv-oof-submission/submission_{n}.csv')\n        test_coded = pd.merge(test_coded, df_[['image_name','target']], on=\"image_name\",how='right')\n        test_coded.rename({'target': f'pred_{n}'}, axis=1, inplace=True)\n    return test_coded\ntest_coded = add_submission_pred(test_coded,5)\ntest_coded.to_csv('test_coded.csv',index=False)\ntest_coded.tail()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def crossValidate(CLF,X=train_coded,X_test=test_coded,FOLDS = 5,SEED = 123,show_roc_curve = False,pesudo_label = False):\n    print(Fore.YELLOW)\n    print('#'*60)\n    model_name = type(CLF).__name__\n    print('#### ',model_name)\n    print('#'*60,Style.RESET_ALL)\n    \n    CV_Score = []\n    Val_preds = []\n    Val_imagenames = []\n    val_targets = []\n    \n    CV_Score_pesudo = []\n    Val_preds_pesudo = []\n    Val_imagenames_pesudo = []\n    val_targets_pesudo = []\n    \n    skf = KFold(n_splits=FOLDS,shuffle=True,random_state=SEED)\n\n    for fold,(idxT,idxV) in enumerate(skf.split(np.arange(15))):\n\n        idxT, idxV = X.tfrecord.isin(idxT), X.tfrecord.isin(idxV)\n\n        X_train_main, y_train_main = X[idxT], X[idxT]\n        X_val_main, y_val_main = X[idxV], X[idxV]\n        print(Fore.MAGENTA)\n        print('#'*60,Style.RESET_ALL)\n        print(Fore.BLUE)\n        print('FOLD : ',fold)\n        print('Train TFrecords : ',X_train_main.tfrecord.unique())\n        print('Validation TFrecords : ',X_val_main.tfrecord.unique())\n        image_names = list(X_val_main['image_name'])\n        \n        X_train = X_train_main.drop(['target','tfrecord'],axis=1).iloc[:,1:]\n        y_train = y_train_main['target']\n        \n        X_val = X_val_main.drop(['target','tfrecord'],axis=1).iloc[:,1:]\n        y_val = y_val_main['target']\n        \n        CLF_pesudo = CLF\n        CLF.fit(X_train, y_train)\n        \n        try:\n            y_train_pred  = CLF.predict_proba(X_train)[:,1]\n        except:\n            y_train_pred = CLF.predict(X_train)\n            \n        print('Train AUC : ', roc_auc_score(y_train,y_train_pred))\n        try:\n            Val_pred  = CLF.predict_proba(X_val)[:,1]\n        except:\n            Val_pred = CLF.predict(X_val)\n            \n        Val_auc = roc_auc_score(y_val,Val_pred)\n        print('Val AUC : ', Val_auc)\n        \n        CV_Score.append(Val_auc)\n        Val_preds.append(Val_pred)\n        Val_imagenames.append(image_names)\n        val_targets.append(list(y_val))\n        \n        if pesudo_label:\n            \n            train2_Pesudo = X_train_main.copy()\n            train2_Pesudo['target_label'] = y_train_main['target']\n            test2_Pesudo = X_val_main.copy()\n            test2_Pesudo['target_label'] = Val_pred\n            \n            test2_Pesudo = test2_Pesudo[ (test2_Pesudo['target_label'] >= 0.99) |  \n                                            (test2_Pesudo['target_label'] >= 0.01)]\n            test2_Pesudo.loc[ test2_Pesudo['target_label']>=0.5, 'target_label' ] = 1\n            test2_Pesudo.loc[ test2_Pesudo['target_label']<0.5, 'target_label' ] = 0\n            \n            print(Fore.CYAN)\n            print('Number of Pesudo Labeled Data added : ',len(test2_Pesudo))\n            print('target_label = 1 : ',len(test2_Pesudo[test2_Pesudo['target_label'] == 1]))\n            print('target_label = 0 : ',len(test2_Pesudo[test2_Pesudo['target_label'] == 0]))\n            \n            train_pesudo = pd.concat([train2_Pesudo,test2_Pesudo],axis=0)\n            \n            X_train_pesudo = train_pesudo.drop(['target_label','target','tfrecord'],axis=1).iloc[:,1:]\n            y_train_pesudo = train_pesudo['target_label']\n            \n            CLF_pesudo.fit(X_train_pesudo, y_train_pesudo)\n            \n            try:\n                y_train_pred_pesudo  = CLF_pesudo.predict_proba(X_train_pesudo)[:,1]\n            except:\n                y_train_pred_pesudo = CLF_pesudo.predict(X_train_pesudo)\n\n            print('Pesudo Train AUC : ', roc_auc_score(y_train_pesudo,y_train_pred_pesudo))\n            try:\n                Val_pred_pesudo  = CLF_pesudo.predict_proba(X_val)[:,1]\n            except:\n                Val_pred_pesudo = CLF_pesudo.predict(X_val)\n            \n            Val_auc_pesudo = roc_auc_score(y_val,Val_pred_pesudo)\n            print('Pesudo Val AUC : ', Val_auc_pesudo)\n            print(Style.RESET_ALL)\n            \n            CV_Score_pesudo.append(Val_auc_pesudo)\n            Val_preds_pesudo.append(Val_pred_pesudo)\n            Val_imagenames_pesudo.append(image_names)\n            val_targets_pesudo.append(list(y_val))\n            \n    valtargets = np.concatenate(val_targets)\n    valpreds = np.concatenate(Val_preds)\n    valimagenames = np.concatenate(Val_imagenames)\n    \n    auc_score = roc_auc_score(valtargets,valpreds)\n    \n    print(Fore.YELLOW)\n    print('#'*60)\n    print('\\nCV(auc_score) : ',auc_score)\n    print(f'Mean CV : {np.mean(CV_Score)} +/- {np.std(CV_Score)}\\n')\n    \n    \n    oof = pd.DataFrame()\n    oof['image_name'] = valimagenames\n    oof['pred'] = valpreds\n    oof['target'] = valtargets\n    \n    \n    ## Test data Prediction\n    Test_imagenames = X_test['image_name']\n    X_test = X_test.iloc[:,1:]\n    try:\n        test_pred = CLF.predict_proba(X_test)[:,1]\n    except:\n        test_pred = CLF.predict(X_test)\n        \n    submission = pd.DataFrame()\n    submission['image_name'] = Test_imagenames\n    submission['target'] = test_pred\n    \n    if pesudo_label:\n        valtargets_pesudo = np.concatenate(val_targets_pesudo)\n        valpreds_pesudo = np.concatenate(Val_preds_pesudo)\n        valimagenames_pesudo = np.concatenate(Val_imagenames_pesudo)\n        \n        auc_score_pesudo = roc_auc_score(valtargets_pesudo,valpreds_pesudo)\n        \n        print('Pesudo CV(auc_score) : ',auc_score_pesudo)\n        print(f'Pesudo Mean CV : {np.mean(CV_Score_pesudo)} +/- {np.std(CV_Score_pesudo)}\\n')\n        \n        oof_pesudo = pd.DataFrame()\n        oof_pesudo['image_name'] = valimagenames_pesudo\n        oof_pesudo['pred'] = valpreds_pesudo\n        oof_pesudo['target'] = valtargets_pesudo\n        \n        ## Test data Prediction (Pesudo)\n        try:\n            test_pred_pesudo = CLF_pesudo.predict_proba(X_test)[:,1]\n        except:\n            test_pred_pesudo = CLF_pesudo.predict(X_test)\n\n        submission_pesudo = pd.DataFrame()\n        submission_pesudo['image_name'] = Test_imagenames\n        submission_pesudo['target'] = test_pred_pesudo\n        \n    \n    print('#'*60,Style.RESET_ALL)\n    if show_roc_curve:\n        fpr, tpr, _ = roc_curve(valtargets,valpreds)\n        if pesudo_label:\n            fpr_p, tpr_p, _ = roc_curve(valtargets_pesudo,valpreds_pesudo)\n            \n        plt.figure()\n        lw = 2\n        if pesudo_label:\n            plt.plot(fpr_p, tpr_p, color='red',\n                 lw=lw, label=f'Pesudo ROC curve (area = {auc_score_pesudo:0.4f})')\n            \n        plt.plot(fpr, tpr, color='darkorange',\n                 lw=lw, label=f'ROC curve (area = {auc_score:0.4f})')\n        \n        plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n        plt.xlim([0.0, 1.0])\n        plt.ylim([0.0, 1.05])\n        plt.xlabel('False Positive Rate')\n        plt.ylabel('True Positive Rate')\n        plt.title(f'ROC Curve by : {model_name}')\n        plt.legend(loc=\"lower right\")\n        plt.show()\n    \n    if pesudo_label:\n        return oof , submission, auc_score, model_name, oof_pesudo, submission_pesudo, auc_score_pesudo, str(model_name) + '_Pesudo'\n    return oof , submission, auc_score, model_name","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# **Classifiers**","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"## LGBMClassifier","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf1 = lgb.LGBMClassifier(max_depth=5, \n                          metric=\"auc\", \n                          n_estimators=100, \n                          num_leaves=5, \n                          boosting_type=\"gbdt\", \n                          learning_rate=0.1, \n                          feature_fraction=0.05, \n                          colsample_bytree=0.1, \n                          bagging_fraction=0.8, \n                          bagging_freq=2, \n                          reg_lambda=0.2)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## LogisticRegression","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf2 = LogisticRegression(\n            C= 1.0,\n            class_weight=None,\n            dual= False,\n            fit_intercept= True,\n            intercept_scaling= 1,\n            l1_ratio= None,\n            max_iter= 100,\n            multi_class= 'auto',\n            n_jobs= None,\n            penalty= 'l2',\n            random_state= None,\n            solver= 'lbfgs',\n            tol= 0.0001,\n            verbose= 0,\n            warm_start= False)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## XGBRegressor","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf3 = XGBRegressor(base_score=0.5, \n                    booster=None, \n                    colsample_bylevel=1,\n                    colsample_bynode=1, \n                    colsample_bytree=0.8, \n                    gamma=1, \n                    gpu_id=-1,\n                    importance_type='gain', \n                    interaction_constraints=None,\n                    learning_rate=0.002, \n                    max_delta_step=0, \n                    max_depth=10,\n                    min_child_weight=1, \n                    missing=None, \n                    monotone_constraints=None,\n                    n_estimators=700, \n                    n_jobs=-1, \n                    nthread=-1, \n                    num_parallel_tree=1,\n                    objective='binary:logistic', \n                    random_state=0,\n                    reg_alpha=0, \n                    reg_lambda=1, \n                    scale_pos_weight=1,\n                    subsample=0.8,\n                    tree_method=None, \n                    validate_parameters=False, \n                    verbosity=None)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## GaussianNB","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf4 = GaussianNB(\n    priors= None, \n    var_smoothing= 1e-09)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## RandomForestClassifier","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf5 = RandomForestClassifier(\n        bootstrap= True,\n        ccp_alpha= 0.0,\n        class_weight= None,\n        criterion= 'gini',\n        max_depth= 5,\n        max_features= 'auto',\n        max_leaf_nodes= 30,\n        max_samples= None,\n        min_impurity_decrease= 0.0,\n        min_impurity_split= None,\n        min_samples_leaf= 2,\n        min_samples_split= 100,\n        min_weight_fraction_leaf= 0.0,\n        n_estimators= 300,\n        n_jobs= None,\n        oob_score= False,\n        random_state= None,\n        verbose= 0,\n        warm_start= False)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## LinearRegression","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# clf6 = LinearRegression()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Lasso","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# clf7 = Lasso()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## ElasticNet","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# clf8 = ElasticNet()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## KNeighborsRegressor","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf9 = KNeighborsRegressor(algorithm= 'auto',\n                            leaf_size= 30,\n                            metric= 'minkowski',\n                            metric_params= None,\n                            n_jobs= None,\n                            n_neighbors= 10,\n                            p= 5,\n                            weights= 'uniform')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"##  DecisionTreeRegressor","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf10 = DecisionTreeRegressor(ccp_alpha= 0.0,\n                             criterion= 'mse',\n                             max_depth= 5,\n                             max_features= 'auto',\n                             max_leaf_nodes= 30,\n                             min_impurity_decrease= 0.0,\n                             min_impurity_split= None,\n                             min_samples_leaf= 2,\n                             min_samples_split= 100,\n                             min_weight_fraction_leaf= 0.0,\n                             presort= 'deprecated',\n                             random_state= SEED,\n                             splitter= 'best')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## GradientBoostingRegressor","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf11 = GradientBoostingRegressor()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## StackingClassifier","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"SCF = StackingClassifier(classifiers=[clf1, clf5], \n                         meta_classifier=clf2,\n                         use_probas=True,\n                         average_probas=True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"if test_pipeline:\n    pipelines = []\n    pipelines.append(('LGBMClassifier', Pipeline([('LGBMClassifier',clf1)])))\n    pipelines.append(('LogisticRegression', Pipeline([('LogisticRegression',clf2)])))\n    pipelines.append(('XGBRegressor', Pipeline([('XGBRegressor',clf3)])))\n    pipelines.append(('GaussianNB', Pipeline([('GaussianNB',clf4)])))\n    pipelines.append(('RandomForestClassifier', Pipeline([('RandomForestClassifier',clf5)])))\n    # pipelines.append(('LinearRegression', Pipeline([('LinearRegression',clf6)])))\n    # pipelines.append(('Lasso', Pipeline([('Lasso', clf7)])))\n    # pipelines.append(('ElasticNet', Pipeline([('ElasticNet', clf8)])))\n    pipelines.append(('KNeighborsRegressor', Pipeline([('KNeighborsRegressor', clf9)])))\n    pipelines.append(('DecisionTreeRegressor', Pipeline([('DecisionTreeRegressor', clf10)])))\n    pipelines.append(('GradientBoostingRegressor', Pipeline([('GradientBoostingRegressor', clf11)])))\n    pipelines.append(('StackingClassifier', Pipeline([('StackingClassifier', SCF)])))\n\n    X_train = train_coded.drop('target',axis=1).iloc[:,1:]\n    y_train = train_coded['target']\n\n    M_name = []\n    M_auc_score = []\n    for name, model in pipelines:\n        cv_results = cross_val_score(model, X_train, y_train, cv=FOLDS, scoring='roc_auc')\n        M_auc_score.append(np.mean(cv_results))\n        M_name.append(name)\n        print(\"%s: %f +/- %f\" % (name, cv_results.mean(), cv_results.std()))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"if test_pipeline:\n    fig, ax = plt.subplots(figsize=(7,7))\n\n    # Save the chart so we can loop through the bars below.\n    bars = ax.bar(\n        x=M_name,\n        height=M_auc_score,\n        tick_label=M_name\n    )\n\n    # Axis formatting.\n    ax.spines['top'].set_visible(False)\n    ax.spines['right'].set_visible(False)\n    ax.spines['left'].set_visible(False)\n    ax.spines['bottom'].set_color('#DDDDDD')\n    ax.tick_params(bottom=False, left=False)\n    ax.set_axisbelow(True)\n    ax.yaxis.grid(True, color='#EEEEEE')\n    ax.xaxis.grid(False)\n\n    bar_color = bars[0].get_facecolor()\n    plt.xticks(rotation=90)\n\n    for bar in bars:\n        ax.text(\n          bar.get_x() + bar.get_width() / 2,\n          bar.get_height() + 0.05,\n          round(bar.get_height(), 4),\n          horizontalalignment='center',\n          color=bar_color,\n          weight='bold'\n        )\n\n    fig.tight_layout()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Setup Parameters With GridSearchCV","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"select_classifier = clf2\nselect_classifier.get_params()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"if Setup_Parameters:\n\n    X_train = train_coded.drop('target',axis=1).iloc[:,1:]\n    y_train = train_coded['target']\n\n    params = dict(\n        C = [0.001, 0.01, 0.1, 1, 10],\n        max_iter = [100,150,50]\n    )\n\n    grid = GridSearchCV(estimator=select_classifier,\n                        param_grid=params, \n                        cv=FOLDS,\n                        scoring='roc_auc',\n                        refit='AUC',\n                        n_jobs = -1)\n\n    grid.fit(X_train, y_train)\n\n    for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n        print(\"AUC : %0.5f +/- %0.5f %r\"\n              % (grid.cv_results_['mean_test_score'][r],\n                 grid.cv_results_['std_test_score'][r] / 2.0,\n                 grid.cv_results_['params'][r]))\n\n    print(f'\\nBest parameters: {grid.best_params_}')\n\n    oof, submission, auc_score, model_name = crossValidate(grid,X=train_coded,X_test=test_coded,FOLDS=FOLDS,SEED=SEED,show_roc_curve=True)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Training and Predictions","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"M_name = []\nM_auc_score = []\nfor CLF in [clf1,clf2,clf4,clf5,clf10,clf11,SCF]:\n    if pesudo_label:\n        oof, submission, auc_score, model_name, \\\n        pesudo_oof, pesudo_submission, pesudo_auc_score, pesudo_model_name = crossValidate(CLF,X=train_coded,\n                                                                                           X_test=test_coded,\n                                                                                           FOLDS=FOLDS,SEED=SEED,\n                                                                                           show_roc_curve=True,\n                                                                                           pesudo_label=True)\n    else:\n        oof, submission, auc_score, model_name = crossValidate(CLF,X=train_coded,X_test=test_coded,FOLDS=FOLDS,SEED=SEED,show_roc_curve=True,pesudo_label=False)\n    \n    oof.to_csv(f'oof_{model_name}_{auc_score:0.4f}.csv',index=False)\n    submission.to_csv(f'submission_{model_name}_{auc_score:0.4f}.csv',index=False)\n    M_name.append(model_name)\n    M_auc_score.append(auc_score)\n    \n    if pesudo_label:\n        pesudo_oof.to_csv(f'Pesudo_oof_{model_name}_{pesudo_auc_score:0.4f}.csv',index=False)\n        pesudo_submission.to_csv(f'Pesudo_submission_{model_name}_{pesudo_auc_score:0.4f}.csv',index=False)\n        M_name.append(pesudo_model_name)\n        M_auc_score.append(pesudo_auc_score)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Classifier Performance","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"fig, ax = plt.subplots(figsize=(15,7))\n\n# Save the chart so we can loop through the bars below.\nbars = ax.bar(\n    x=M_name,\n    height=M_auc_score,\n    tick_label=M_name\n)\n\n# Axis formatting.\nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nax.spines['left'].set_visible(False)\nax.spines['bottom'].set_color('#DDDDDD')\nax.tick_params(bottom=False, left=False)\nax.set_axisbelow(True)\nax.yaxis.grid(True, color='#EEEEEE')\nax.xaxis.grid(False)\n\nbar_color = bars[0].get_facecolor()\nplt.xticks(rotation=90)\n\nfor bar in bars:\n    ax.text(\n      bar.get_x() + bar.get_width() / 2,\n      bar.get_height() + 0.05,\n      round(bar.get_height(), 5),\n      horizontalalignment='center',\n      color=bar_color,\n      weight='bold'\n    )\n\nfig.tight_layout()","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}