Batching rule for torch.squeeze(tensor) (#47632)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/47632
This one is fun because we have to be careful not to squeeze out any of
the batch dims (it is the dims of the per-example tensor that are being squeezed).
Test Plan: - new tests
Reviewed By: anjali411
Differential Revision: D24859022
Pulled By: zou3519
fbshipit-source-id: 8adbd80963081efb683f62ea074a286a10da288f
