Reliability - log heartbeat checkpoint during broadcast push notification dispatching

[f-w]

During broadcast push notification dispatching, main request may fail, for example due to node failure. In order to re-try the dispatching, a precursor is being able to detect a main request has failed.

This can be achieved by having the main request update notification record at a fixed interval, say 1 min. If the notification is in sending state and last update timestamp is much older than the interval, it indicates the main request has failed.

This is part of reliability improvements to achieve node failure resilience. See #36 for a full description.