> each Kafka partition was represented by a separate SQLite database

What does the programming interface look like for this? I believe you'd only be able to use SQL to query each partition (database) individually, unless there's another layer orchestrating SQL over the partitions. I wonder if you just end up with connectors again.

Additional databases can be added to a SQLite connection using:

  ATTACH 'pathname.db' AS prefix

https://www.sqlite.org/lang_attach.html