Comment 7 for bug 1641153

I started working on this some days ago and I've got what I consider to be the workers ready.
Now I have to implement the worker manager, which is the one that will create the threads and feed them with the tracks to analyze.

It was a bit complicated to understand correctly how do the slots/signals operate in a multithreaded scenario, and combine them correctly with mutex, waitconditions and the likes.

I've done it this way:

- Manager creates the worker and thread and assigns them (todo)
- worker creates the analyzers (each worker needs their own instances) when starting (not in the constructor) sends a signal to the manager to ask for a track and waits on a waitcondition.
- The manager receives the message and calls the worker directly feeding it one track (todo. The manager will know the worker because the worker itself tells how it is with the message).
- The worker wakes up and processes the track. It sends signals to the manager to inform about the progress.
- The manager receives the progress signals, and sends more signals for UI update.
- The worker finishes analyzing the track and sends again the message to the manager that it needs a new track.
- If the manager has no more tracks to analyze, it calls the worker method to stop.
- The worker then sends a signal to the manager to indicate that it is finishing, and also sends a signal to the thread and they both clean up automatically.

I.e. threads will be alive only while they are required.

The scenario of prioritizing tracks (i.e. analysis of the player tracks) goes as follow:
- If the manager has allocated max number of threads, take one of the existing workers and tell it to pause. Pause means wait on a waitcondition similarly to what they do to get a new track. The granularity is each processed buffer.
- Then, it allocates a new worker, which is then configured for the foreground analysis type (i.e. ensuring waveform analysis is on). This worker will work similarly to other workers and when it finishes, it will ask too for a new track.
- If there is no more priority tracks to analyze, the manager will tell the priority worker to stop and cleanup, and then it will tell the paused worker to continue.

Most of the manager part is still on todo, but the parts on the worker that the manager will use are already there.

Hope this will be good enough and not too complicate to understand either (hey.. it wasn't that easy right now anyway)