For now the pause policy is that every 200 tokenized chars, if half a frametime has passed, it "yields".
We might consider allowing the customization of the pause policy.

I think it's fine; very few people are going to customize this. If anyone wants to, they can complain, and we can put in a config variable, but I really don't think it's necessary.

The yielding behavior is not implemented with coroutines because of performance concerns around creating them. I might test if this is a valid concern.

I honestly think it's fine; it's clear enough, and not a huge amount of code.

Only tokenization coroutines are allowed to tokenize whole lines.
Highlighter:get_line only tokenizes the first 200 characters, plus anything else it can in the allowed time frame.
This is done because that function can be called multiple times per frame, which could tank the performance. We might consider adding a parameter to allow resuming the tokenization.

Seems reasonable.

When the tokenizer pauses, the remaining text uses the incomplete token type. Suggestions for a different name are welcome.

I'm good with this.

We sadly can't resume the tokenization of a line from where it changed, and need to restart the tokenization from scratch. This is because the change itself could match with a rule that applies before it.

Honestly, that's fine, and reasonable for now. If it becomes an issue with people who really want to edit really long lines, while tokenizing them, we can always revisit.

I'm going to do a bit more testing, and if it looks good, I'll merge.

Yeah, tested; looks good. Merging.