[llvm] AMDGPU: Materialize bitwise not of inline immediates (PR #95960)

[Matt Arsenault via llvm-commits]

arsenm wrote:

> The first 2 changes I thought to avoid the VOPD regression didn't work. Checking if it's a post-RA run and moving the VOPD pass both still see the same regression

The VOPD lit regression was from now handling the reversed FP immediate case. We were always interfering with VOPD here 

https://github.com/llvm/llvm-project/pull/95960