input is used. When stream_Read was used alone, it read 1316 bytes from
the RTP input add added garbage upto DUMP_BLOCKSIZE. With stream_Peek it
works as expected and also makes the whole demuxdump faster.
Signed-off-by: JP Dinger <jpd@videolan.org>

DEMUX_GET_PTS_DELAY must not fail for an access_demux.

 The aout core use a hack to avoid useless memcpy: the buffer in which
to mix is the same as the one in the first active input fifo.

Contrary to swscale(?), those plugins call avutil directly

It would seem to be significantly faster that way.

This was unstable.

We don't need it.

We don't want users to not press this button just because there is an other one afterwards.

Signed-off-by: Felix Paul Kühne <fkuehne@videolan.org>

alloca() was not used anyway on OSX and BSD, due to smaller stack sizes,
and we can't assume a default stack size anyway

I expect the performance loss to be minimal, but worth the code
simplification anyway (i didn't benchmark)

aout_BufferAlloc() is moved into its own function in a .c file instead
of being a macro

Since there is now 2 types of allocations (HEAP and NONE), make
i_alloc_type a boolean (true = HEAP alloc, false = NO alloc)

make aout_BufferFree() a static inline function in the same process.
Prototype doesn't change since the provided buffer doesn't need to be
set to NULL (I checked all the callers)

Signed-off-by: Rémi Duraffort <ivoire@videolan.org>