It may not be uncommon for the input timestamps to experience some jitter
around the 'perfect time'. As such, instead of regularly adding and dropping
samples, optionally allow for some tolerance in a more relaxed approach.
API: GstAudioRate:tolerance