Finish queuing system

[AlexandreHutton]

Queue system is half-implemented; I ran into problems with multiprocessing spawning new Flask instances that blocked the original instance from running correctly.

Currently: submitting a job will run it immediately. The model isn't resource-intensive, the caps on the individual submissions are enough to limit accidental misuse, and we don't expect heavy use, so it isn't much of a problem.

Due to the half-implementation, there's a bug in the event that two users submit a job at roughly the same time: there's a small window of time that might result in the first submission being run twice. The results will only be returned to the submitter, but the second job will sit in the queue until the next user submits a job, and then that job will sit in the queue until the next submission, etc.

To fix:

• Have a background process regularly check if there are any submissions in the queue; run the next one if so.
• Have each job submission trigger the background process instead of running directly.