Draft: Resolve "Final quick muting phase after many lost frames incorrectly implemented in PhaseECU"

• Related issues:
• Requested reviewers: Norvell

Reason why this change is needed

*Bug in final muting towards zero of phase ecu noise component beta

Description of the change

Change noise component muting factor (beta_mute) update from per sub-band to the correct per frame update.

Affected operating points

Very long bursty FER (larger than 10 lost frames in a row) with HQ-MDCT core active

1. Produced bitstream is BE compared to previous state. (EVS for now)
2. Produced bitstream is non-BE but it is fully backwards compatible for decoding. Decoded output may differ. (IVAS)

• For output difference, use BE or non-BE. Additionally, amount of difference can be presented.
  • IVAS decoder results
    • RANDOM FER with burst shorter than 10 frames, BE CI tests do not have long bursts
    • LONG burst FER . beta noise component now attenuated per frame (slower) and not per sub band (very quick).
      • note: Noise component 'beta' is only a minor part of the PLC , as the major energy contribution is coming from the phase evolved 'alpha' component.
  • EVS_mono decoder results : BE for now
• long burst FER file will be added with this MR filename: testv/FEC_bl10to69in70.g192 (9 bursts over 630 frames with length in the region 10 to 69 )
  • ( separate MR to be created for the addition/update of CI tests )

Related to ticket #1180