[llvm] [X86][BF16] Improve vectorization of BF16 (PR #88486)

[Simon Pilgrim via llvm-commits]

https://github.com/RKSimon approved this pull request.

LGTM - although I think eventually it'd be nice to get bf16->f32 "zext+shift" out of X86 and handle it in generic legalization

https://github.com/llvm/llvm-project/pull/88486