Can you clarify what you mean in terms of OSM features import? I ask b/c, in my understanding, the only ways to import OSM data are to download it from GeoFabrik or similar, or to pull it down from OverpassAPI. I've seen your Bars&Pubs notebook a few times (sorry, I realized I hadn't liked before now), and the main challenges I've had in applying in are that 1) it relies on having bbox definitions, and 2) that it hard-codes in the overpass API queries. Overpass API is really hard to understand and generalize as a novice. I read this notebook as being a quick-and-dirty tutorial for getting common spatial data types into Observable. Maybe my general reaction is that -- bringing OSM into this equation may not entirely fit this scope? (I comment with full love for this notebook and great admiration for the many works you've shared!!!)
I think it's really cool that you wish to show how to connect to Apache Parquet databases and to render them into Observable. Compared to the other spatial data formats presented here, however, it seems rather involved. My reading of Kyle Barron's notebook linked above (from a novice reader) is that it uses Apache Arrow to interpret the parquet file format as data, and then undertakes a series of data interpretation steps to transform these data into a spatial dataset that can be rendered into the browser. I am very interested to see your take on this... and similarly very interested to see Parquet-formatted data sources in the wild that could be interpreted! Depending on what you want to achieve, it may make sense to limit this notebook to the more common types you've already demonstrated.. and just to publish as it ;) I've been looking for this sort of quick and dirty intro for a while, so you're already filling a gap!