[llvm] [AMDGPU][Codegen] Don't incorrectly shrink `S_LOAD_DWORDX{16,8}_IMM` during remat (PR #225209)

[Igor Wodiany via llvm-commits]

https://github.com/IgWod commented:

Thanks! LGTM, but I'll let someone more familiar with this code to approve it.

https://github.com/llvm/llvm-project/pull/225209