When using autodetection, AlsaData.device wasn't properly initialized
with NULL.  This broke autodetection randomly.

The variables mod_mikModInitiated and mod_mikModInitError were used to
control lazy initialization, but they are superfluous now.

The "mod" decoder plugin was being initialized lazily, but was
deinitialized unconditionally.  That led to segmentation faults.
Convert mod_initMikMod() to be the global module initialization
method.  The MPD core should care about lazy initialization.

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

An empty chunk may happen when it only contains a tag, but no PCM
data.  Don't call playAudio() then.

ffmpeg_tag() deleted the tag when ffmpeg_helper() returned success.
The return value was interpreted incorrectly, it should return the tag
on success.

Cleaning up artifacts from a merge gone wrong.  Also remove the
ringbuf library, which is not being used.

Non-local songs used to have no tags.  If the decoder sends us a tag,
we should incorporate it into the song struct.  This way, clients can
always show the correct song name (if provided by the server).

Always remember which song is currently being sent to the audio
device.

Rewinding the stream here is not useful, but may consume valuable
resources (and time).

The try_decode() method may have read some data from the stream, which
is now lost.  To make this data available to other methods, get it
back by rewinding the input stream after each try_decode() invocation.

The ogg and wavpack plugins did this manually and inconsistently; this
code can now be removed.

If a chunk contains a tag, send it to the audio output device.  Few
output plugins support this, e.g. shout has support for sending tags.

If the source chunk has a tag, merge it into the destination chunk.
The source chunk gets deleted after that, and this is our last chance
to grab the tag.

decoder_read() checks the decoder command.  Without this patch, the
ffmpeg plugin could become unresponsive.

Ogg and ffmpeg detection was disabled when the stream was not
seekable, because the detection was too expensive.  Since the curl
input stream can now rewind the stream cheaply, we can re-enable
detection on streams.

During codec detection, the beginning of the stream is consumed.  This
is a common operation, which takes a lot of time when handling remote
resources.  To optimize this, remember the first 64 kB of a stream.
This way, we can rewind the stream without actually fetching the start
of the stream again.

Provide an API for submitting additional tags from the stream.

Avoid while(true) loops, and convert them to a loop with a proper
condition.

Since the aac and mod plugins have told MPD that they cannot seek, MPD
will never send a SEEK command to them.  Removed the SEEK comand
checks from both plugins.

Don't pass the "seekable" flag with every decoder_data() invocation.
Since that flag won't change within the file, it is enough to pass it
to decoder_initialized() once per file.

Each music chunk can now carry a tag object.  Decoder plugins which
support it (e.g. oggvorbis) may use this to inject decoded tags into
their output.

These two functions will care about memory allocation and deallocation
in the future.

Add doxygen compatible comments.

The "volatile" keyword doesn't help here, because we have proper
memory barriers, but it disables some optimizations.  Remove it.

The name "num_chunks" expresses the meaning of the variable better.

Rename all variables and struct members.

Don't bother to pass these values as parameters to tail_chunk().

Last music_pipe rename patch: renamed the global variable (singleton).

Replace all direct music_pipe struct accesses with wrapper functions.
The compiled machine code is the same, but this way, we can change
struct internals more easily.

Eliminate the deprecated utils.h memory allocation functions.

.. and rename dc.audioFormat to dc.in_audio_format.  The music pipe
does not need to know the audio format, and its former "audioFormat"
property indicated the format of the most recently added chunk, which
might be confusing when you are reading the oldest chunks.

Don't make tail_chunk() calculate the frame size again.

Rename all functions to the new prefix.

.. and rename ob_chunk to struct music_chunk.

No CamelCase in the file name.  The output_buffer struct is going to
be renamed to music_pipe.  There are so many buffer levels in MPD, and
calling this one "output buffer" is wrong, because it's not the last
buffer before the music reaches the output devices.

Commit 1a4a3e1f moved decoders into a static array, but failed to
enable those plugins who did not have an init() method at all.
This patch corrects the "enabled" check.

The strings were constant, but the pointers weren't.  C syntax is
somewhat tricky..

All decoder_plugin structs are initialized at compile time, and must
never change.