This mostly affects the Pulse output plugin which needs to "cork" the
stream (closes #278).

When switching to another song manually, the player checks if the
decoder is already decoding that song; if so, it will attempt to reuse
it by seeking it to the new position.  That however fails if the
decoder is not seekable (e.g. a radio stream) which leaves the user
unable to switch to that song with the bogus error message "Not
seekable".

Fixes deadlock bug.  Closes #210

Closes #203

This is another instance of the kludge we already have a few lines
above that.  It works around a deadlock/race condition revealed by
commit 98a7c62d

The output plugin shall decide whether to insert silence or do nothing
at all.  The ALSA output plugin has already implemented this.
Inserting silence is not necessary or helpful for some plugins, and
may even hurt them (e.g. "recorder").

Eliminates unnecessary mutex locks.

When the decoder is still starting up while we handle a SEEK, finish
the "player SEEK" immediately and re-enter the player loop, being able
to handle commands (and even cancel the pending seek).

This is the first part in a series of patches to solve the "blocking
input blocks decoder, blocks player, blocks the main thread" problem.
There are many other blocking code locations left, and the main thread
isn't non-blocking either because it waits for "seeking" to become
false.

The STOP, EXIT and CLOSE_AUDIO commands are not finished here; they
are propagated to PlayerControl::RunThread() where the
outputs.Cancel() call will be done again.

Allows signalling a failed seek, and replaces several redundant
command checks after the ProcessCommand() call.

Decouple this function from player command execution.

Use one single std::lock_guard in Run(), and replace the manual
Lock()/Unlock() calls.

Eliminate more unlock/lock transitions, making the code more robust.

Regression from commit 0772e571

Reduces main thread contention.  Avoids blocking the main thread in
"status" commands.