Make the iff demuxer send the whole audio chunk to the decoder as a
single packet, move stereo interleaving from the iff demuxer to the
decoder, and introduce an 8svx_raw decoder which performs
stereo interleaving.

This is required for handling stereo data correctly, indeed samples
are stored like:
LLLLLL....RRRRRR

that is all left samples are at the beginning of the chunk, all right
samples at the end, so it is necessary to store and process the whole
buffer in order to decode each frame. Thus the decoder needs all the
audio chunk before it can return interleaved data.

Fix decoding of files 8svx_exp.iff and 8svx_fib.iff, fix trac issue #169.