[llvm] [AMDGPU] Optimize away v_readfirstlane_b32 on SGPR input (PR #151033)

[Matt Arsenault via llvm-commits]

https://github.com/arsenm commented:

Needs IR tests, but my gut feeling is this is solving a problem you shouldn't need to be solving 

https://github.com/llvm/llvm-project/pull/151033