original implementation didn't correctly check the size of the target buffer in the required size check
original implementation had (imo) ambiguous naming for things making it harder to understand
original implementation didn't resize the sampling buffer if it ever was too small (it would very quickly be too small! 512 samples @16k would not fit into a 960 sized buffer upsampled to 48k)
This tweak comes with regressions however:
VoiceManager._resampleBuffer is no longer readonly because we need to be able to resize it
AudioResample.Resample no longer accepts a channel count argument, although that was not used
This solves the following problems:
This tweak comes with regressions however: