[llvm] [AMDGPU] Add getCmpSelInstrCost override to re-enable SimplifyCFG speculation for vector types (PR #208043)

[Matt Arsenault via llvm-commits]

================
@@ -5,8 +5,8 @@ define amdgpu_kernel void @divergent_cycle_1(i32 %a, i32 %b, i32 %c) {
 ; CHECK: CYCLES ASSUMED DIVERGENT:
 ; CHECK:   depth=1: entries(R P) S Q
 ; CHECK: CYCLES WITH DIVERGENT EXIT:
-; CHECK:   depth=1: entries(R P) S Q
 ; CHECK:   depth=2: entries(S P) Q
+; CHECK:   depth=1: entries(R P) S Q
----------------
arsenm wrote:

Unrelated change 

https://github.com/llvm/llvm-project/pull/208043