[discard] Attempt fast-shutdown regardless of active workers

Worker lifetimes may extend beyond the lifetime of their associated
document indefinitely, which can result in the retention of a
renderer for which all frames have been discarded.

This CL updates the fast-shutdown path for renderers hosting
discarded RenderFrameHosts to disregard workers, aligning with the
behavior of the existing implementation.

This CL also shortens the retry interval to avoid renderers staying
alive longer than they should following a discard operation. This
is important as discard is typically triggered in response to
acute contention of system resources and discard should release
associated resources as fast as possible.

The deficiency of the new vs the old discard implementation is seen
in a 5% regression in total memory at the 50th percentile, driven
by an uptick in reported renderer processes.

Bug: 382145604
Change-Id: I7ede0881de241a9a070b6e73493dd8eb99ff5eec
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/6071382
Code-Coverage: findit-for-me@appspot.gserviceaccount.com <findit-for-me@appspot.gserviceaccount.com>
Commit-Queue: Thomas Lukaszewicz <tluk@chromium.org>
Reviewed-by: Rakina Zata Amni <rakina@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1392382}