I now wish that I had engaged with this idea that was very similar to litestream that I had about a year and half ago. I always thought SQLite just needed a distribution layer to be extremely effective as a distributed database of sorts. Its flat file architecture means its easy to provision, restore and backup. SQLite also has incremental snapshotting and re-producible WAL logs that can be used to do incremental backups, restores, writes etc. It just needs a "frontend" to handle those bits. Latency has gotten to the point where you can replicate a database by its continued snapshots (which is, on a high level, what litestream appears to be doing) being propagated out to object / blob storage. You could even achieve brute force consensus with this approach if you ran it in a truly distributed way (though RAFT is probably more efficient).
Reason I didn't do this? I thought to myself - why in the world in 2020 would someone choose to use SQLite at scale instead of something like Firebase, Spanner, Fauna, or even Postgres? So after I did an initial prototype (long gone, never pushed it to GitHub) I just felt like...there was no appetite for it.
Now I regret!
Just a long winded way of saying, congrats! This is awesome! Thanks for doing exactly what I wanted to do but didn't have the guts to follow through with.