[non-BE] Resolve "old_inp_16k_fx potentially wrong scale in stereo_mdct_core_enc_fx()"

• Related issues:
• Requested reviewers:

Reason why this change is needed

• In !2553 (merged) many rescale operations to Q0 and Q-1 in the encoder preprocessing were removed, allowing a variable Q for 12k8, 16Khz and related signals. Many of them controlled by Q_new. But it was missed to remove a rescale to Q0 and back to Q_new-1 for MDCT processing, only one instance of rescaling to Q0 was removed, resulting in the 16KHz signal in Q0 format while other function would still assume it is Q_new-1.

Description of the change

• Remove all rescalings in the context of MDCT encoder processing of the 16 KHz signal, and adding an additional Q parameter to stereo_mdct_core_enc_fx() and ivas_mdct_core_whitening_enc_fx() to be able to pass through the Q_new-1 value instead of rescaling.
• Also the format documentation was corrected for core_signal_analysis_high_bitrate_ivas_fx() (part of that signal path).

Affected operating points

• All operating points using 16KHz signal path and MDCT modes.

Closes #2549 (closed)