reference backups by unique IDs (not url or name)

The idea is to store the UUID in the backup files, and use it for the name of the cache dir.
Then the correct cache should always be identifyable.
Backup --name option would be stored with the backup files and in the cache.
Time of last backups, URLs etc. could be cached for integrity, informational or convenience reasons.

If duplicity full/incremental/restore is called only with the --name (or an --ID) option, duplicity could use the cached URL(s) and settings.

* The cache name would not contain a hash of the url, and stay valid if url changes.
* Named backups could be renamed without breaking older copies.
...