rqlite creator here.
Yes, you still need the data at every node. rqlite is not about edge-compute specifically (though many solutions have come along after rqlite was created to address that need, I guess I should have responded to that demand much earlier instead :-)). One of my primary goals with rqlite has always been ease-of-deployment and ease-of-use. I wanted to build a quality distributed database that was both really reliable and super easy-to-use.
Check out the FAQ, which might answer some more of your questions: https://rqlite.io/docs/faq/