Currently, the db session is initialized once at startup time (implicitly, when you import rida).

It is then re-used forever in rida.scheduler.main.

If there is a network failure and we lose connectivity to the db for a brief moment, there is no way to recover other than to restart the scheduler.

Instead of using a global shared session, the session should be created and destroyed every time a new message is received off the incoming queue in the WorkerThread.