Add support for non-affine batch norm with float stats and half inputs (#22750)
Summary:
This PR creates support for non-affine batch norm with float running estimates and half inputs.
Changed were made similar to https://github.com/pytorch/pytorch/issues/16735.
I couldn't find a specific test for `SyncBatchNorm`, so I used [this code](https://gist.github.com/ptrblck/ab45bfcde6df55ac28a7be18531f4718) to test it.
cc ngimel
Pull Request resolved: https://github.com/pytorch/pytorch/pull/22750
Differential Revision: D17119965
Pulled By: ezyang
fbshipit-source-id: 2e8c5d63fc3c636b8a1338c43c9c101a0f5e9b22