The core issue doesn't include "non-object".

I suppose you added it to account for explicit object parameters.

Is that really enough? No need to restrict it to an explicit object parameter, but any amount of non-object parameters is OK? Is there wording elsewhere that make these cases ill-formed, and thus saying "non-object" is enough?

How about the function type specified below? Does it need a revision, too?

I think the paper misrepresents the status quo? So tihs seems like a "plausible merge".

A conversion function with an explicit this parameter seems plausible, but there must not be other parameters. Further cleanup (if any) should be in a separate editorial or core issue.