bfloat16: vectorized clamp, clamp_min and clmap_max (#35082)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/35082
Test Plan: Imported from OSS
Differential Revision: D20721148
Pulled By: ngimel
fbshipit-source-id: 949b66e28bfc6049a891fced9ea308131b3675c6
