Turns out, I got it to work! I've created a demo here: https://kevmo314.github.io/appendable/ and observe that it queries GitHub Pages, so nothing but static hosting while fetching a fraction of the underlying data (ie proportional to the result set, not the input set). The hope is that for certain datasets, the Appendable can be used to host the database with static file hosting instead of the traditional, more expensive database service routes.
The demo right now is pretty bare bones as I was pretty excited to get it working, but I outlined an aspirational roadmap in the linked repository that includes more features.
Thanks for checking it out!