@oblomov Sorry, I assumed you'd be busy, and put it on a back-burner. There will be a write-up, but this development is only preipherally related to it. This technique was one of four that I drafted up trying to make a quick-and-dirty in-web database system mathematically neater. Turns out that it makes it somewhat more powerful, so much so that two important use cases of it generalise in ways that exclude each other. But I've still got three left, and even if they all should fail — which seems unlikely —, well, I could just spec up the quick-and-dirty thing. Not ideal, but it has working code.
I have several meetings today, but I'm still planning a write-up soon, and I'd still like your feed-back. If you'd like to not wait, I could offer you the not-yet-fully-fleshed-out raw list of potential features when I get to home (probably after dark in CET, alas).