Add segmental SNR to pytest results

In https://www.3gpp.org/ftp/tsg_sa/WG4_CODEC/3GPP_SA4_AHOC_MTGs/SA4_Audio/Docs/SA4aA240038.zip, an SNR limit is proposed. " For decoding, the output synthesis of FX decoder (FL enc+ FX dec) compared to FL decoder (FL enc + FL dec) should be within the limit of segmental SNR (per frame and then averaged) > [60 dB], and MLD < 5. "

It would be useful to add a segmental SNR to the pytests, e.g. by implementing this option in the Python comparison tool cmp_pcm.py and including this number in the daily reports.