[codegen][arm64] Avoid one use of PatchingAssembler

Creating a PatchingAssembler has significant overhead, including a
dynamic allocation for the assembler buffer implementation. In the case
of {Assembler::bind} we just need it to overwrite a machine word. Hence
avoid creating the PatchingAssembler for this trivial work and just use
Memcpy directly.

R=jkummerow@chromium.org

Change-Id: I83510cfd7ebdb0d0c378df548b442eabf3727aeb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2668827Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#72523}

src/codegen/arm64/assembler-arm64.cc

@@ -574,8 +574,7 @@ void Assembler::bind(Label* label) {
       // Internal references do not get patched to an instruction but directly
       // to an address.
       internal_reference_positions_.push_back(linkoffset);
-      PatchingAssembler patcher(options(), reinterpret_cast<byte*>(link), 2);
-      patcher.dc64(reinterpret_cast<uintptr_t>(pc_));
+      base::Memcpy(link, &pc_, kSystemPointerSize);
     } else {
       link->SetImmPCOffsetTarget(options(),
                                  reinterpret_cast<Instruction*>(pc_));