The scheduler logic

Once we have a set of parallel tasks, and a set of nodes that are available to us, we should distribute those tasks to the nodes in a way that minimizes the execution time, since in the end all that matters is to get the job done as fast as possible.

This is something that is not well implemented yet. We do almost no real scheduling, as the system is today. This was a lower priority, and didn't get the attention it needs.