SCReAMv2: Replace non-linear backoff scaling with spec backoff

Replace the stateful UseNonLinearBackoffScale state machine in
ScreamV2::ScaleBackoff with the congestion backoff mechanisms from
draft-ietf-ccwg-rfc8298bis-screamv2-02 (controlled via UseSpecBackoff,
enabled by default):

1. Inhibit delay-based backoff (is_virtual_ce) when queue_delay_avg is
   decreasing relative to queue_delay_avg_prev_ (updated once per
   min(rtt, ReactionInterval())), and only set is_virtual_ce at a
   reaction interval when !is_loss && !is_ce.
2. Attenuate backoff by BackoffScaleLowTargetRate (0.25) when
   target_rate_ < AckedBitrateMargin (0.8) * received_rate_ to prevent
   reference window undershoot when RTT spikes (Section 4.5.3).
3. Scale backoff linearly across reactions per RTT via
   max(1.0, rtt / ReactionInterval()).

Bug: webrtc:447037083
Change-Id: I029fe66442db4be35e86400ac9a52639122237ad
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/510185
Commit-Queue: Per Kjellander <perkj@webrtc.org>
Reviewed-by: Rasmus Brandt <brandtr@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#48992}