[m76] Ignore FrameHostMsg_SetNeedsOcclusionTracking in inactive frames.

A frame might get deleted (e.g. via its parent/ancestors) while some IPCs
targeting the frame are still in flights.  In this case, the IPC should
be dropped.  In particular, in this case it is not valid to continue and
assume that GetProxyToParent will return non-null value - the parent
might also be getting deleted and we might have already proactively
removed parent's proxies (see https://crbug.com/950625) and are just
waiting for parent's unload handler to finish.

Note that other 3 places where bad_message::RFH_NO_PROXY_TO_PARENT might
get triggered are already checking |is_active()| first.

TBR=lukasza@chromium.org

(cherry picked from commit 0e067b8d07e4fa74c1bc0982b5cdc5c490674311)

Bug: 972566
Change-Id: If7b1bff8082621621a9f9afc901aedd5e5a259b1
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1651885
Commit-Queue: Łukasz Anforowicz <lukasza@chromium.org>
Reviewed-by: Alex Moshchuk <alexmos@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#667758}
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1655971
Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org>
Cr-Commit-Position: refs/branch-heads/3809@{#255}
Cr-Branched-From: d82dec1a818f378c464ba307ddd9c92133eac355-refs/heads/master@{#665002}

• content/browser/frame_host/render_frame_host_impl.cc[diff]
• content/browser/frame_host/render_frame_host_manager.h[diff]