Use duplicacy as backend

[RubenKelevra]

Duplicacy allows for space efficient backups with multiple machines to multiple locations, while doing compression and deduplication over all blocks submitted to a storage.

As it deduplicates locally, it would save a lot of bandwidth as no blocks already existing are sent to the storage.

Only drawback: Only uncompressed tar archives make sense here - so we need to decompress the archives or change the way HA creates them.

As it's written in Go, there are no dependencies, and you just need to ship a binary.

It can do parallel backups and cleanups of backup locations without locking each other's processes. So perfect for nightly backups.

And yes it supports encryption for the storage via password and on top encryption via private/public GPG keys for the data itself on top of that and it can also make sure two backup locations are in sync for our backup. So we could save a backup to a NAS and a backup to Google Drive for example and keep them in sync.