Document checkpoint and logs file/directory layout

[alexanderkjeldaas]

In the latest hackage downtime description, backups were 2 weeks old.

If acid-state could write to lots of small files, then doing backups at 60 second intervals would be a trivial docker commit && docker push command.

It is unclear what controls when a new log file is created, the layout of the acid-state directory, and what effect creating a snapshot has.

Answers to this should be added to the FAQ or the documentation, specifically how incremental backup can be done and if incremental backup that is limited to file granularity (such as a docker image layer) works well or not with acid-state.