[Fix] Fix potential bug in ThreeNN

[yezhen17]

Currently this bug won't be triggered in the repo. However, when the indices calculated by ThreeNN participate in backwarding, there will be an error message: RuntimeError: Expected isFloatingType(grads[i].scalar_type()) to be true, but got false.

Please refer to facebookresearch/votenet#72 and https://discuss.pytorch.org/t/custom-autograd-function-with-int-tensor-input/65907.

[codecov]

Codecov Report

Merging #377 (d48f400) into master (22c47ac) will decrease coverage by 0.00%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master     #377      +/-   ##
==========================================
- Coverage   49.55%   49.54%   -0.01%     
==========================================
  Files         177      177              
  Lines       11834    11835       +1     
  Branches     1844     1844              
==========================================
  Hits         5864     5864              
- Misses       5610     5611       +1     
  Partials      360      360              

Flag	Coverage Δ
unittests	49.54% <0.00%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
mmdet3d/ops/interpolate/three_nn.py	50.00% <0.00%> (-2.64%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 22c47ac...d48f400. Read the comment docs.