parser plugins should parse top-level unknown and void elements

[bantic]

The Section parser first infers a section type for a given element before attempting to parse its child nodes, and it only runs its parser plugins against the child nodes. This means that HTML with root-level elements do not get run through the plugins. This happens (in slightly different ways) with both void elements and elements with childNodes:

• parse('<video>...</video>') — plugins skip <video>
• parse('<video>') — plugins skip video

[kevinansfield]

I've just run into this too. I'd like to be able to parse <figure><img/><figcaption>Foo</figcaption></figure> into our image card in order to extract the caption for the payload but the <figure> element never hits the parser plugins, only the <img>.

I have a similar problem where a <hr /> element doesn't get passed to the parser plugins so we're unable to convert it to our divider card. Is this the same problem or does that stem from somewhere else?