By introducing the NativeInputTrack in bug 1702646, now we have an AudioSegement containing de-interleaved audio input data, which will be shared to AudioInputProcessing as well. We could use this as the input for AudioInputProcessing::PacketizeAndProcess so it doesn't need to deinterleave the raw input data itself.

In bug 1725810, we will have a copy of the raw interleaved audio input data, for doing AudioInputProcessing::PacketizeAndProcess. With this improvement, this copy could be saved.

In bug 1741959, we will pass an AudioSegment to PacketizeAndProcess. The deinterleaved data in AudioSegment will be converted to interleaved, do mixing, and then deinterleaved. We should be able to skip doing interleave since the data given is deinterleaved. If we need to do upmixing or downmixing, AudioChannelsDownMix and AudioChannelsUpMix should be able to work with the deinterleaved data.

Set severity to S3 since its not a critical blocking issue