[llvm] [AMDGPU] Generate COPY for each use-constraint instead of constraining the register class (PR #182104)

[Matt Arsenault via llvm-commits]

================

----------------
arsenm wrote:

I think you don't need to actually copy. You can just try a constrainRegClass(getCommonSubClass())

https://github.com/llvm/llvm-project/pull/182104