Sorry for the question but how behind are the LLMs in terms of quality for something like this?

I can't really answer that because I haven't actually tried to use an LLM on any part of the process. The vast majority of the process is semantic markup using (x)html and proofreading. The markup process could, I guess, use an LLM, but most of it is already automated using regex and linting.

Does it use any automation?

My bro-in-law supported his family as a freelance editor for years while my sister was doing the "maternity leave" thing so I know there's a non-trivial amount of work that goes into book editing. Cutting out some of that human labor seems like a good thing for a volunteer project.

there is quite a lot of automated changes using standard ebooks open source tools package

the vast majority of textual tooling is regex-galore, but there is also automated epub tooling in there too