[llvm] [AMDGPU] Constrain AV->VReg if we do not exceed RP thresholds (PR #150086)

[Matt Arsenault via llvm-commits]

https://github.com/arsenm commented:

I don't really see the point of doing this. All the downstream register allocator code is free to re-inflate these right back to the AV_*. If we do not exceed the pressure threshold, we can always just remove the A registers from the allocation order by reserving them. The scheduler should not be actively trying to increase restrictions just in case. If you're changing the instruction type, you can adjust the register class directly as part of that transform 

https://github.com/llvm/llvm-project/pull/150086