[llvm] [AMDGPU] Don't fix the scavenge slot at offset 0 (PR #79136)

[Matt Arsenault via llvm-commits]

https://github.com/arsenm commented:

This means that we need to fully handle access by incrementing SP and decrementing after. That was broken at some point. Do we still hit the PEI logic to keep the scavenging slot as close to the incoming SP as possible? 

https://github.com/llvm/llvm-project/pull/79136