action=parse&prop=headhtml is supposed to give "parsed <head> of the page". Instead, it returns something like
<!DOCTYPE html> <html ...> <head> ... </head> <body ...>
when I would have just expected the <head> element and its contents.
I'm not too sure what "parsed" means in this context, either. What would an unparsed <head> look like?
This should either be fixed (which would clearly be a breaking change), or the documentation corrected to match the actual output. It's not clear from the comments on T20608 what the intended output was at the outset.