Yes, I did this for self-contained reports in maybe 2014. All images referenced (containing diagrams) were embedded as data URIs. Restrictions are AFAIK more picky now, though so YMMV in 2021.

Cool project! It is kind of interesting that the link is the content, not sure it’s always useful, but for twitter like short form content perhaps?

Well webpages themselves including links, the embedding would need to be recursive.

not the entire content, but a hash id is pretty common https://en.wikipedia.org/wiki/Magnet_URI_scheme