Implement a more robust queueing mechanism

[blarghmatey]

This should have a more robust queueing mechanism so that stuck jobs can be placed in a lower priority queue to prevent traffic jams due to a single failing request.

At the least there should be a timeout for inactivity so that the task at the head of the queue gets cycled to the back to allow subsequent requests to be processed.

[pdpinch]

Closed by #9