When the audio outputs are closed, also clear the audio format.  If we
don't do this, every call to audio_output_all_update() will open the
device, even if it's meant to be paused.

Don't allow reopening an audio device after pause with
audio_format==NULL, force the caller to provide the audio_format each
time.

Instead of passing individual buffers to audio_output_all_play(), pass
music_chunk objects.  Append all those chunks asynchronously to a
music_pipe instance.  All output threads may then read chunks from
this pipe.  This reduces MPD's internal latency by an order of
magnitude.

Sometimes, audio_output_update() isn't called for the second device
when the first one has succeeded.  The patch
"audio_output_all_update() returns bool" broke it, because the boolean
evaluation ended after the first "true".

audio_output_all_update() returns true when there is at least open
output device which is open.

Do error handling with GError instead of aborting with g_error().

time() is not a monotonic timer, and MPD might get confused by clock
skews.  clock_gettime() provides a monotonic clock, but is not
portable to non-POSIX systems (i.e. Windows).  This patch uses GLib's
GTimer API, which aims to be portable.

When all outputs are disabled, MPD printed only a meaningless message
"problems opening audio device", although it didn't attempt to open a
device.

After initialization, audio_outputs is always non-NULL.  Don't check
that.

Don't include output_api.h in output_internal.h.  This change requires
adding missing includes in several sources.

When MPD explicitly starts playing, ignore the "REOPEN_AFTER" timeout.
This timeout was useful when MPD attempted to reopen a failed device
over and over, but it confuses users when they explicitly tell MPD to
start playing, while MPD insists to wait for the 10 seconds to pass.

audio_output_all_finished() returns bool, not int.

Renamed functions and variables.

Moved code which deals with all audio outputs at once into a separate
library.

This function isn't used anymore.

Added audio_output_get(), audio_output_find().

audio_output_config_count() returns the number of audio outputs in the
configuration file.  It is only used by initAudioDriver().  The public
function audio_output_count() now returns audioOutputArraySize.

On some platforms, g_free() must be used for memory allocated by
GLib.  This patch intends to correct a lot of occurrences, but is
probably not complete.

Those have been superseded by the new legacy configuration code.

All config_get_block_*() functions should accept constant config_param
pointers.

Renamed functions, types, variables.

Only include headers which are really needed.  os_compat.h aimed to
make MPD easily portable, but was never actually made portable.

Replaced myFgets() with fgets() + g_strchomp().

Don't use NOTIFY_INITIALIZER to initialize audio_output_client_notify.

Nobody should call playAudio() with an empty chunk.  Add some
assertions on that.

Since open() and play() close the device on error, we can simply check
audio_output.open instead of audio_output.result after a call.

When one of several output devices failed, MPD tried to reopen it
quite often, wasting a lot of resources.  This patch adds a delay:
wait 10 seconds before retrying.  This might be changed to exponential
delays later, but for now, it makes the problem go away.

Moved code from syncAudioDeviceStates() to audio_output_update().

Stopping an audio output device without cancelling its buffer doesn't
make sense.  Combine the two operations, which saves several cancel
calls.

Don't return 0/-1 on success/error, but true/false.  Instead of int,
use bool for storing flags.

Instead of implementing another loop which calls audio_output_close()
on all configured devices, simply call closeAudioDevice().