Starts using celery queues

[noirbizarre]

This PR introduce Celery routing and routing usage.

This create by default 3 queues:

• high for high priority tasks (ie. search indexation)
• default for common tasks
• low for slow tasks (ie. harvest tasks by example)

This allows to start workers like this:

celery -A udata.worker -Q high worker
celery -A udata.worker -Q high,default worker
celery -A udata.worker -Q high,default,low worker

and will ensure that high priority tasks are always handled and low priority tasks (slow tasks) won't flood the default queue.

If no -Q parameter is set, the worker still consume all the known queues.