Using datasette with Jupyter Server Proxy.
This fork is an experiment to see if https://github.com/simonw/datasette/issues/394 solves the various asset problems.
datasette port is assigned dynamically, although a known port could be specified if datasette is used to provide an API. (But in a notebook context that probably doesn't make sense? You could just as easilu call directly on the SQLite3 db.)