[wasm-simd][ia32] Small optimization for i64x2.abs

We no longer need unique register for src, since I64x2Abs
macro-assembler handles the case when dst == src (by using the scratch).
This is equivalent to calling the VisitRRSimd helper.

Bug: v8:11190
Change-Id: Ibabb6da970e983ca2c9326057a6dc43cc7b7c435
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2750725Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#73385}

src/compiler/backend/ia32/instruction-selector-ia32.cc

@@ -3173,14 +3173,7 @@ void InstructionSelector::VisitI64x2GeS(Node* node) {
 }
 
 void InstructionSelector::VisitI64x2Abs(Node* node) {
-  IA32OperandGenerator g(this);
-  if (CpuFeatures::IsSupported(AVX)) {
-    Emit(kIA32I64x2Abs, g.DefineAsRegister(node),
-         g.UseUniqueRegister(node->InputAt(0)));
-  } else {
-    Emit(kIA32I64x2Abs, g.DefineSameAsFirst(node),
-         g.UseRegister(node->InputAt(0)));
-  }
+  VisitRRSimd(this, node, kIA32I64x2Abs, kIA32I64x2Abs);
 }
 
 // static