Is it OK to do a review and not give a fully working implementation, i.e. with "some assembly required"?
Absolutely, see @Mat's answer.
Or can I expect the OP to fill in the blanks if I show them the way?
If you've given constructive comments, then congratulations, you've made a good contribution. If the OPauthor doesn't know how to implement your suggestions, they have to research. That's normal, as we're also not teach-you-how-to-code.com
What to do when user insists you spoon feed them?
Although, just as you are in no way obliged to spoon feed, the OPauthor is in no way obliged to accept (or even upvote) your otherwisepossibly brilliant answer. It can also happen that another reviewer swoops in, credits you for your excellent suggestions (hopefully), and provides an example implementation, and snatches that much coveted check mark. It's part of the game.
What I do depends on my mood and on the question. If the problem is interesting enough, or if I just really, really want that check mark NOW, then I might do it. Other times I disengage early and simply ignore the requests for spoon feeding. There's not a whole lot we can do about such requests.