Keep track of whether there is data being played, and don't call
AudioOutput::Drain() after Cancel() has been called already.

In this state, we can't make any progress.

Closes https://github.com/MusicPlayerDaemon/MPD/issues/1084

Drain only if it was requested explicitly.

It is useless, because we're always checking pa_stream_is_corked().

Preparing for a variant of TimerEvent with coarse 1-second
granularity, but cheaper (with a timer wheel).

Fixes Windows compatibility.

To make things simpler, AllocatedString is now a non-template class.

Found with cppcoreguidelines-interfaces-global-init
Signed-off-by: Rosen Penev <rosenp@gmail.com>

Fixes #994 - moveoutput: new AudioOutputControl created from copyMoving an output to a partition is now done via MultipleOutputs::AddCopy(),using a new AudioOutputControl constructor. Tags and always_on settings willpersist when moving outputs between partitions.

Similar to commits 1686f4e8 and
30a5dd26

This allows canceling the blocking method LockWaitWriteAvailable(),
and thus allows breaking free of misbehaving ALSA drivers, avoiding a
MPD lockup.

Closes https://github.com/MusicPlayerDaemon/MPD/issues/966

This allows interrupting the output thread (for some plugins which
implement this method).  This way, operations can be canceled
properly, instead of waiting for some external entity.

Move the InternalPlay() call and the wake_cond.wait() call into the
`case Command::NONE` and revert all `continue` statements to a simple
`break`.

"DEFAULT" is a bad name - all it says is that it's the default value,
but it doesn't say what it means.  The name NOTICE mimics the syslog
level.

Found with -Wdouble-promotion
Signed-off-by: Rosen Penev <rosenp@gmail.com>

Our AudioObjectGetPropertyDataT() wrapper throws exception on error,
and calling it from OSXOutput::Disable() can cause MPD crash due to
std::terminate().

Closes https://github.com/MusicPlayerDaemon/MPD/issues/932

According to https://developer.android.com/ndk/guides/audio/opensl/android-extensions

This feature was mentioned in https://github.com/MusicPlayerDaemon/MPD/issues/922

Wait until there is data in the ring buffer.

Wait until there is some data; don't let our render callback be
invoked without any data.

We shouldn't restart the AudioUnit while the ring buffer is empty, or
else our render callback may emit noise.

Closes https://github.com/MusicPlayerDaemon/MPD/issues/771

This will keep track of AudioOutputUnitStart() and
AudioOutputUnitStop().  This will provide some separation between "not
(yet) (re)started" and "paused".

This shall be done by Close(), which will be called automatically
after an error.