In fact, the rates may be identical. Resampling might be used only to
compensate accidental drift. (We could also remove channel and format
checks.)

Reported-by: Aliz Hammond

FT_Face.family_name can be NULL for embedded fonts.

See also http://www.freetype.org/freetype2/docs/reference/ft2-base_interface.html#FT_FaceRec

This is really crappy, but it's as good as I can make it with such a
dumb interface.

This enable more uses of optimized format conversion plugins.
(Currently the generic format converter hides indirect conversions.)

This should fix problems when remixing is required but the FL32
format is not involved, as well as decoding on non-FPU platforms.

This also disables (or rather avoids) remixing in A52 and DTS decoders.
I do not really see the point in using the A52 downmixer anyway.

this worked out of pure luck for the past 6 years

Drop audio rather than abort.

Close #6200

Ref #6895
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

The latency needs to be returned also when not (yet) running to start
playback on time (the audio output core inserts zeroes accordingly).

Also, then there is no point to using PCM status, as only one single
parameter is used. This simplifies the code and works around a bug
(probably in ALSA plugin framework) whereby delay was always zero
when using the ALSA<->PulseAudio I/O plugin.