Parallelize feed fetching and score calculation

[bergie]

Now we only fetch one feed at a time and import. Feed importing would pass through a lot faster if we had multiple worker processes handling it.

[indeyets]

but proper implementation of this would require some kind of queueing solution. do we have a bug for that?

We need a queue of tasks, with per-task locking and unlocking (if unsuccessful). then we can have multi-process dequeuer, which is started by cron each minute

[bergie]

Queue implementation would indeed make this easier. Would you like to propose how it would work?