[llvm] [AMDGPU] Use s_cmovk_i32 instead of s_cselect_b32 when applicable (PR #135232)

[Ryan Buchner via llvm-commits]

bababuck wrote:

Made changes for the above comments. I left a couple responses to the comments above as there was some wording I didn't quite understand.

https://github.com/llvm/llvm-project/pull/135232