[llvm] [AMDGPU] Swap V_CNDMASK operands to shrink it into VOP2 (PR #135162)

[Ana Mihajlovic via llvm-commits]

mihajlovicana wrote:

> It would probably be easier to handle this as a pre-selection combine

i need to check source modifiers to determine if the swap is beneficial 

https://github.com/llvm/llvm-project/pull/135162