Surface synchronization: Use the frame rect in pixels for physical backing size

Prior to this CL, RenderWidgetHostViewChildFrame converted the frame rect
in DIP to physical pixels for GetPhysicalBackingSize. The frame rect in DIP
is computed using ScaleToEnclosingRect which can change the size of the rect
depending on its position. This occasionally results in a surface invariants
violation when surface synchronization is turned on.

The solution is to use the frame rect in physical pixels from the parent
and pass that along in RWHVChildFrame::GetPhysicalSize().

Bug: 672962
Cq-Include-Trybots: master.tryserver.chromium.linux:linux_site_isolation
Change-Id: Ic1ec779b8330e77c04fe3a3c0a73b90d57b4beaf
Reviewed-on: https://chromium-review.googlesource.com/753832
Commit-Queue: Fady Samuel <[email protected]>
Reviewed-by: Mitsuru Oshima <[email protected]>
Reviewed-by: Antoine Labour <[email protected]>
Cr-Commit-Position: refs/heads/master@{#517294}
6 files changed