[non-BE][split-non-BE] Resolve "Bug in tcx ltp synth filtering"

• Related issues:#856 (closed)
• Requested reviewers: @bayers tagging additional people to check different affected formats, SBA: @jtorr , ISM, stereo: @vaclav, MASA: @pihlajakuja

Reason why this change is needed

• There is an error in function tcx_ltp_synth_filter_11_unequal_pitch(), the length of the decreasing gain filtering is wrong and should be corrected to length instead of length + L

Description of the change

• Describe what is done.

Affected operating points

• Not BE for all modes when TCX core coding is used see https://forge.3gpp.org/rep/ivas-codec-pc/ivas-codec/-/pipelines/24107/test_report. However, differences are very small. As an example differences for the condition stereo at 32kbps 32kHz out, I get the following information from the ssnrcd tool:

channel: 0 
    samples compared                              : 640000
    number of different samples                   : 123798
    max. abs. difference sample                   : 9.16e-05 (threshold: 6.10e-05, 16 bits res.)
    overall RMS value of difference               : -96.12 dB (threshold: -101.10 dB, 16 bits res.)
    average segmental SNR value of difference     : 73.18 dB (320 samples per segment)
    cepstral distortion                           : 0.02 dB (320 samples per segment)
channel: 1 
    samples compared                              : 640000
    number of different samples                   : 163271
    max. abs. difference sample                   : 1.83e-04 (threshold: 6.10e-05, 16 bits res.)
    overall RMS value of difference               : -94.00 dB (threshold: -101.10 dB, 16 bits res.)
    average segmental SNR value of difference     : 71.18 dB (320 samples per segment)
    cepstral distortion                           : 0.03 dB (320 samples per segment)

Related to #856 (closed)