Wrong entity encoding in story HTML

GitLab fails to create the issue fork for some reason so here is a patch.

Turns out Html::load() removes everything outside the body so this is not what we need here.
The root problem seems to be that DOMDocument::loadHTML() does not detect the encoding correctly. Forcing it like this works but does not feel very clean:

    $success = @$dom->loadHTML('<?xml encoding="utf-8" 

' . $html);
?>

Using the HTML5 library seems to work correctly (it is what Html::load() uses internally).

I can confirm patch 3 is solving the encoding issue (for me with german Umlauts..)
provided the patch as an MR

Automatically closed - issue fixed for 2 weeks with no activity.