I have to say I'm similarly not really sure I'm getting it, and I've been following litestream pretty closely over the years (not an expert but certainly familiar).
I think what we're getting here is a way to just spin up a local shell / app and run arbitrary queries from any point in time over the network without having to sync the full prod database. I guess with LiteFS you would have to do this, or pre-plan to do this, it's not totally on-demand.
Or said another way, do things locally as though in prod without having to ssh to prod and do it there (if you even can, I guess if 'prod' is just s3 you can't really do this anyway so it's an entirely new capability).
@benbjohnson is this right? I humbly suggest adding a tl;dr of the main takeaway up top of the post to clarify. Love your work on litestream, thanks for what you do!
Your second paragraph is correct, with the added proviso that you can build features in your own application tooling to take advantage of the same capability.
One reason you're not getting such a clear usage statement at the top of this post is, it's an early feature for a general-purpose capability. I think we might rather get other people's takes about what it's most useful for? There are some obvious use cases, like the one you just identified.