I’m not sure why they have a limit, I’m sure there is a good performance reason.
I think you are expected to mix documents in buckets to an extent. With 2.0 you aren’t expected to only look up documents by ID (KEY). You can look them up on indexed fields.
You use the map function to handle different documents in a given bucket. So in map.js for the view you handle the different documents appropriately.