Set up job queue so work is not lost

In order to read usage and perform transactions without losing data due to application shutdown or other faults, the service should implement a persistent job queue. Jobs may include:

• Performing the initial reading
• Scheduling readings to happen periodically
• Adding credits at IAA PoP start
• Zeroing out credits at IAA PoP end

---

Security considerations

The River module is a new dependency, but it will use the postgresql database we are already intending to deploy with the service.

Implementation sketch

I intend to use River to implement the queue.