PCScan: Make gcc compile avx2 and sse3 intrinsics

The CL also try to slightly improve throughput of the hot path in the
loop by using vmovdqa instead of vmovapd and avoiding bypass delays
between integer and floating-point execution units.

Bug: 11297512
Change-Id: Ie2aa830fe3785b09d2d16fb6a8e7a639b0ff16a4
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2674004


Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Anton Bikineev <bikineev@chromium.org>
Cr-Commit-Position: refs/heads/master@{#850539}