For float to int conversions and vice versa the MacroAssembler on x64
was using a mix of AVX and non-AVX instructions.
This CL fixes that to consistently use AVX if available.

R=jkummerow@chromium.org

Change-Id: I3aecda9b99881254b24949ced5bed870fdc2a754
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1678361Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62387}