Xiaolong

Hi Linfang, The NPCS baseline result looks correct to me, but the nonlinear one doesn't. While the ransac-based nonlinear solver doesn't always guarantee a better estimation, it shouldn't be such...

Hi Linfang, I found there is actually a step missing in the cleaned [ransac](https://github.com/dragonlong/articulated-pose/blob/267b70a41e9fbbc3ab7a2b50baa5b5ea07e278d2/evaluation/parallel_ancsh_pose.py#L31) function, after each iteration, the best score should be updated as well, or this will result...