Allow a safe Node recovery from crashes #656

This targets moments which, if experiencing a panic, don't allow the Node fully recover because in the new session some information is missing. Lost during the panic, and even though it should rightly be possible to navigate the Node to gets its bearings back by providing information stored in the DB, not always we succeed to keep the DB up to date with all information at once, and thus most of the time, this issue concerns some kind of data inconsistency or simply a DB corruption.

Note that these events are either difficult or even impossible to predict. It depends on the cause of the crash. We can fight panics within our application if we know the conditions leading to it, but we have no luck with crashes of the host the Node is running on.

This is a parent card. There will probably be multiple cards as its children (their count may extend after a deeper analysis) that try to address these issues.