I'm planning to continue porting forward to subsequent versions of SpiderMonkey until we have ES6 modules. (They seem to be nearly finished.) I wouldn't spend time rewriting the module system only to have it replaced by ES6 modules later on...
Yes, my opinion is this should have been done from the beginning; it's strange that it works for properties but not for methods. I support this but only if it turns out we can do it without breaking existing code and without imposing a performance burden. I would probably not do the transformation on the JS side for performance reasons. If I were able to find out the C++ code where it would need to be done, and outline an approach, would that make it comfortable enough for you to work on it?
I have some vague ideas about this that need fleshing out. What from jsgtk do you consider made it the most fun and cool? It would be good to know where we could make the most gains first.Philip