No subject

[bogus at does.not.exist.com]

planar and packed, by setting in/out to the first sample of each
channel, and supports only 6 channels.

In libavfilter we need to support up to 8 channels, and we need to
pass the information packed/planar when initing the converting context
(or alternatively in av_audio_convert2).

I wonder if this is a good occasion for sketching the layout of
libavresample, audioconvert could be the first piece to be moved
there.

Anyway I'll try hard to see a simpler solution, not involving deep
changes to the current API.
-- 
FFmpeg = Fantastic and Faithless Merciless Pitiless EntanGlement