[llvm] [AMDGPU] Emit b32 movs if (a)v_mov_b64_pseudo dest vgprs are misaligned (PR #160547)

[Matt Arsenault via llvm-commits]

================
@@ -93,3 +93,12 @@ body: |
   bb.0:
     $vgpr0_vgpr1 = V_MOV_B64_PSEUDO 4575657222473777152, implicit $exec
 ...
+
+# GCN-LABEL: name: v_mov_b64_misalign
+# GCN: $vgpr5 = V_MOV_B32_e32 0, implicit $exec, implicit-def $vgpr5_vgpr6
+# GCN: $vgpr6 = V_MOV_B32_e32 0, implicit $exec, implicit-def $vgpr5_vgpr6
+name: v_mov_b64_misalign
+body: |
+  bb.0:
+    $vgpr5_vgpr6 = AV_MOV_B64_IMM_PSEUDO 0, implicit $exec
+...
----------------
arsenm wrote:

Also test with unaligned AGPR tuple.

Also test unaligned cases for V_MOV_B64_PSEUDO. This is currently in the wrong test file, there's another test for the av_mov_b64 pseudos 

https://github.com/llvm/llvm-project/pull/160547