XML namespaces were a bad idea, and trying to cast HTML into XML turned out to be a bad idea, too. Trust me, I was there when it happened

Among others, the requirement for well-formedness of the whole document made it really hard to have a valid page from 2 sources in any setting that was a little more complex. Especially third party providers can easily make documents break easily.

Right. So any tool acting in the interest of the end user, especially the browser, chose to parse “invalid” markup, too, and rightly so

It could be, but this isn't a theoretical argument. Adopting XML in HTML has been tried through XHTML, with less then stellar results. I reckon there won't be another shot short of inventing a new document format altogether.