[llvm] [X86] Improve transform for add-like nodes to `add` (PR #83691)

[Matt Arsenault via llvm-commits]

================
@@ -426,48 +426,44 @@ body: |
     ; SI: liveins: $sgpr0_sgpr1
     ; SI-NEXT: {{  $}}
     ; SI-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+    ; SI-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY [[COPY]].sub0
+    ; SI-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY [[COPY]].sub1
     ; SI-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 2147483647
-    ; SI-NEXT: [[COPY1:%[0-9]+]]:sreg_32_xm0 = COPY [[COPY]].sub1
----------------
arsenm wrote:

It makes no sense that there are any amdgpu test diffs here 

https://github.com/llvm/llvm-project/pull/83691