jupyter-book: Documentation build hangs indefinitely for pdfhtml on master

No the key thing here is that you only need to register a node if they will actually reach the builder. Since tableofcontents nodes are all now replaced/removed in a transform (that is always applied, as opposed to the current post-transform that is not), they can never reach the builder. Hence they are not registered anywhere

@chrisjsewell @choldgraf I heard from Takeshi (on the dev community channel) and we may be able to have a look at adjusting this upstream in sphinx.

This seems to be an issue during the html -> pdf step using pyppeteer as

jb build docs --builder=custom --custom-builder=singlehtml

the sphinx build for singlehtml seems to be working as well as the html builder.

However when I run the builder=pdfhtml the html that is used in the _build/html folder doesn’t load due to some error so must be an issue in the javascript loading. The left hand index is missing when I open _build/html/intro.html when building with pdfhtml

@choldgraf does the pdfhtml use the singlehtml builder but at the page level? Scrap that it builds into a single file