[roll] Roll fuchsia [audio_core] Refactor AudioCore to use the new Clock types

This change is an almost entirely a mechanical renaming of the following
types:

AudioClock            => Clock
FakeAudioClock        => FakeClock
AudioClockFactory     => AudioCoreClockFactory
FakeAudioClockFactory => FakeAudioCoreClockFactory

But there are some more subtle changes:

* In audio_driver.cc, a few lines of clock recovery logic have are now
  subsumed by the new RecoveredClock class.

* In mix_stage.{h,cc}, a ClockSynchronizer was added to each input edge
  and is now used in place of AudioClock static methods during clock
  reconciliation.

* In base_renderer.cc, most of the clock duplicatation logic has been
  removed since Clocks are passed around by shared_ptr (unlike
  AudioClocks, which were passed around by unique_ptr).

* In mix_stage_unittest.cc, a few small additions were needed because of
  the "minor changed" discussed in fxrev.dev/680507's CL description.

* While most unit tests that previously used FakeAudioClocks now use
  SyntheticClocks, some tests in audio_{capturer,renderer}_unittest.cc
  need to use real clocks. Specifically, every test that calls
  GetReferenceClock needs to use a clock that is backed by a real
  zx::clock. SyntheticClocks are not, so we must use RealClocks.

Original-Bug: 87651
Original-Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/677853
Original-Revision: 5f861eea0d20283b0f3046f30186bcda483909d5
GitOrigin-RevId: 0c55c7e6c5543e8b0be09092bcc796bcaea1e859
Change-Id: I37e208cfc7352d8a68385c4788a3de1ab4e431cc