[llvm] [AMDGPU] Move InferAddressSpacesPass to middle end optimization pipeline (PR #138604)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,60 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S -O1 -mtriple=amdgcn-amd-amdhsa %s -o - | FileCheck --check-prefix=INFER %s
+; RUN: opt -S -O2 -mtriple=amdgcn-amd-amdhsa %s -o - | FileCheck --check-prefix=INFER %s
+; RUN: opt -S -O3 -mtriple=amdgcn-amd-amdhsa %s -o - | FileCheck --check-prefix=INFER %s
+; RUN: opt -S -passes="lto<O1>" -mtriple=amdgcn-amd-amdhsa %s -o - | FileCheck --check-prefix=NO-INFER %s
+; RUN: opt -S -passes="lto<O2>" -mtriple=amdgcn-amd-amdhsa %s -o - | FileCheck --check-prefix=INFER %s
+; RUN: opt -S -passes="lto<O3>" -mtriple=amdgcn-amd-amdhsa %s -o - | FileCheck --check-prefix=INFER %s
----------------
arsenm wrote:

that's how it's printed and most tests use it 

https://github.com/llvm/llvm-project/pull/138604