Getting rough timestamp information for a backup without a cluster

Currently we need a live cluster to get timestamp information from a backup. This can be a problem if we are trying to identify old backups that don't have clusters running anymore. We can't tell the difference between a backup that has only recent data that we're keeping around for a short time after a cluster is decommissioned and a backup that has been long abandoned but needs to get cleaned up. We are going to be adding an approximate time difference between the snapshot boundaries and restorable points, relative to the latest log version in the backup, but without a timestamp from that latest log version we cannot get that into real time, which means we cannot tell how long the backup has been stagnating for. Could we add some time information from the file timestamp to the `backup describe` output, so that we can convert these backup times into a rough estimate of real time?

Start by locating the backup describe command and the code that reads backup file metadata. Trace how snapshot boundaries and restorable points are presented, then determine how file timestamps should appear as an approximate real-time estimate. Done means backup describe reports usable rough time information without requiring a live cluster.

Written by the indexing model from the issue text.