[llvm] [AMDGPU] Move kernarg preload logic to separate pass (PR #130434)

[Matt Arsenault via llvm-commits]

================
@@ -523,6 +525,8 @@
 ; GCN-O1-OPTS-NEXT:      Scalarize Masked Memory Intrinsics
 ; GCN-O1-OPTS-NEXT:      Expand reduction intrinsics
 ; GCN-O1-OPTS-NEXT:      Early CSE
+; GCN-O1-OPTS-NEXT:    AMDGPU Preload Kernel Arguments
----------------
arsenm wrote:

I'd expect this to benefit from maximally canonical IR and run a bit earlier

https://github.com/llvm/llvm-project/pull/130434