also return AVERROR(EINVAL) instead of -1

Put the new raw GSM demuxer in its own file.
Fixes raw GSM demuxing.

calculate it from c->code_size instead.

This is an easy alternative to setting bit_rate. This patch also selects the
closest bit_rate to the requested one rather than requiring an exact value.

This should never happen, but the check avoids a divide-by-zero.

This requires some workarounds in the WAV muxer and demuxer. We need to write
the correct bits_per_coded_sample and block_align in the muxer. In the
demuxer, we cannot rely on the bits_per_coded_sample value, so we use the bit
rate and sample rate to determine the value.

This avoids having the decoder rely on AVCodecContext.bit_rate, which is not
required to be set by the user for decoding according to our API.

This also allows for not having a decoder close function.

Allows for checking output buffer size and simplification of decoding loop.

Signed-off-by: Martin Storsjö <martin@martin.st>

This symbol name clashes with a symbol in gnutls, if linking
statically to that library.
Signed-off-by: Martin Storsjö <martin@martin.st>

The names used in the libdirac/libschroedinger wrappers are long enough as-is.
Bloating them with unnecessary prefixes makes them even more unwieldy.

Also check output buffer size instead of truncating output.

bits_per_coded_sample should be 8.
block_align is calculated incorrectly, but it is not needed anyway.
packet pts should be calculated in samples.
packet duration can be set.