The playback queue is now kept in sync with the playback queue of the
remote jellyfin instance the manager is controlling.
Some additional guards were added in place in the shuffle and playlist
algorithm, since the situation can occur where the now playing index
falls outside of the playing playlist. This happens because when the
an playlist update is received, we need to do another HTTP request
before we know which items are in the queue, while the now playing index
has been updated.
This is a not-optimal way to fix that, but it works well enough for now
and a better solution can be implemented later. (Hello, person in the
future reading the git blame output!)