coVarsOfType is almost surely wrong
coVarsOfType currently looks in the types of coercion variables, but not in the kinds of type variables. It's hard to imagine how this behavior is correct. Either we should look in both kinds, or in neither.
coVarsOfType currently looks in the types of coercion variables, but not in the kinds of type variables. It's hard to imagine how this behavior is correct. Either we should look in both kinds, or in neither.