Currently, all RQ jobs are distributed between 3 queues:

1. periodic - handles periodic jobs stated in schedule.py.
2. schemas - handles refresh_schemas jobs.
3. default - for all other jobs.

A couple of issues that this causes:

1. Since different types of jobs go to default, it's hard to get an understanding of the current distribution of jobs. A solution to this might be to split default into job-specific queues, then we could just observe the queue size.
2. Within periodic, the refresh_queries job needs to have the highest priority.
3. Within default, the send_email jobs needs to have the highest priority.