Support spellchecking HTML

[tbroadley]

Inspired by #123.

This would probably involve using rehype to parse the HTML and rehype-retext to convert it into a format that the spellchecker knows how to spellcheck. I expect this would be relatively easy!

It might involve using old versions of these packages that are compatible with the old version of retext that Spellchecker uses. Or upgrading Spellchecker to use new versions of retext and retext plugins.