Support BFloat16 for binary logical operators on CUDA (#42485)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/42485
Test Plan: Imported from OSS
Reviewed By: ngimel
Differential Revision: D23684423
Pulled By: mruberry
fbshipit-source-id: edc2b46b726361d4c8bf8a4bf4e4a09197b20428
