Event Bus - add persistence and error queue

[ianco]

Add a persistence mechanism to the event queue. Currently events are stored in an in-memory queue, which is lost if the agent re-starts, and is not available between agent instances in a scaled environment (such as OpenShift). The event_bus "wait_for_event" feature won't necessarily work in a scaled environment (can't ensure that a response will return to the same instance of the agent that initiated the request).

Add an error queue, in case an error occurs processing an event, and add a facility to report and/or retry events that generated an error.