[llvm] [AMDGPU] Teach iterative schedulers about IGLP (PR #134953)

[Austin Kerbow via llvm-commits]

================
@@ -2698,6 +2698,13 @@ bool IGroupLPDAGMutation::initIGLPOpt(SUnit &SU) {
 
 namespace llvm {
 
+namespace AMDGPU {
+bool isIGLPMutationOnly(unsigned Opcode) {
----------------
kerbowa wrote:

This function probably fits better in SIInstrInfo.

https://github.com/llvm/llvm-project/pull/134953