Crosscut

I'm interrupting the normal flow of notes for a quick update: Yesterday, I finished implementing restricted structural editing. I went down the wrong path there for a while, making the implementation too complicated. Which is why it took a few weeks. But I found a better approach, and it ended up being a rather simple change.

The current editing capabilities are extremely basic though. I reduced the number of tests (which I think is a decent proxy for feature complexity) from around 90 to less than 60. The result is not good enough for real-world use, but definitely sufficient for making progress in other areas.

I will keep improving editing, as needed. But right now, my focus is on implementing bindings. The lack of which blocks making sensible use of recursion, which will in turn open up a whole new world of problems and opportunities.