[💡FEATURE REQUEST]: Current impl of the worker_watcher queue container may violate Pop <-> Push operation under some circumstances

[rustatian]

Edit: Fixed in the 2.4.0 for the worker's container based on the channel. As the target of this ticket, we need to create a new container described above.

Under some circumstances (TTL, supervisor), when using custom plugins, the order of Pop and Push operations may be violating. When Pool takes the Worker, another worker, which waits on the channel mutex might be pushed into the channel. So, the Pop <-> Push relation might be broken and the worker won't be able to return a response.
Another issue here is if the user used TTL without requests, the killed processes will be on channel mutex:

func (v *Vec) Push(w worker.BaseProcess) {
	v.workers <- w
}

We need to do the following:

1. Remove the duplicated slice with processes pointers from the worker_watcher.
2. Implement the Queue data structure which should support the following operations (all O(1)): Pop, Push, Lish (O(n)), Remove, Destroy. This data structure should also support the allocate_timeouts.
3. Integrate this data structure with worker_watcher.
4. Remove operation should be used to remove worker by its PID from the Queue without stacking dead processed on the mutex. The dead worker might be at any position in the Queue.
5. Rethink List() operation. We don't need to return a full process, but only the Memory, CPU, Execs, Time parts.

[rustatian]

fixed by: roadrunner-server/sdk#36