Fixes for Pytorch 1.7.1+cu110 and CUDA 11.2

[3846chs]

I forked from https://github.com/MatthewHowe/DCNv2 which fixes to be compatible with [torch 1.7 / 1.8 with cuda 10 / 11] This folder structure is slightly different from the original, which can cause errors in several projects using DCNv2. So, I made the folder structure the same as the original. Here is it: https://github.com/3846chs/DCNv2.git

My environment: torch 1.7.1+cu110 cuda 11.2 RTX 3090