[llvm] [AMDGPU][GlobalISel] Fix BITOP3 selecting B16 opcode for 32-bit vector types (PR #217048)

[Shilei Tian via llvm-commits]

https://github.com/shiltian commented:

Seems readable to me, though I thought the size mismatch is handled by legalizer instead of selection. I'm not very familiar with GISel, so I'll leave it to someone else.

https://github.com/llvm/llvm-project/pull/217048