[llvm] [AMDGPU] Treat image_msaa_load as a sampler operation (PR #141726)

[David Stuttard via llvm-commits]

================
@@ -57,9 +57,9 @@ body: |
     ; GFX11: S_WAITCNT 1015
     ; GFX1150-NEXT: S_WAITCNT 1015
     ; GFX12-NEXT: S_WAIT_SAMPLECNT 0
-    ; GCN-NEXT: renamable $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_MSAA_LOAD_V4_V2_gfx11 killed renamable $vgpr4_vgpr5, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 4, 7, -1, 0, 0, -1, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
+    ; GCN-NEXT: renamable $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_LOAD_V4_V2_gfx11 killed renamable $vgpr4_vgpr5, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 4, 7, -1, 0, 0, -1, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
----------------
dstutt wrote:

This change is because MSAA is now classed as a sample, so no longer demonstrates the issue being tested?

https://github.com/llvm/llvm-project/pull/141726