Edits that make substantive changes to other people's code (rather in questions or answers) should be rejected. If they're doing something more than fixing an obvious typo or reformatting code blocks that are difficult to read, they probably aren't a valid edit.
Even if you know the language and agree that their proposed "fix" is a "better" way of doing things, you still shouldn't approve the edit. That is an abuse of the editing system. You don't need to be a domain expert to judge an edit that is useful or inappropriate. Suggestions on how to improve someone's code should be posted as a comment to the post, or perhaps as an entirely separate answer (depending on the nature and scope of the corrections). We don't want to use edits to put words into people's mouths, especially when it comes to code.
I generally reject these types of edits as an "invalid edit", implying that they should have been posted as a comment instead. If you feel more comfortable rejecting them as a "radical change", that's fine, too. It really doesn't matter which rejection reason you pick, the point is that we shouldn't be allowing these through.
On the one hand they add interesting and useful information and rejecting them seems counterproductive as the user might not take the time to create a separate answer after the rejection (or the user is anonymous in which case I would have thought s/he doesn't even get notified)
This is a potential concern, but I don't think it overrides the cardinal rule not to make significant modifications to other people's code.
If you want to educate these users (which is, of course, the point of a suggested edit system, as "training wheels" for future editors), then you could reject their edit with a custom reason. Something like:
Making substantive changes to other people's code via edits is discouraged. Please either leave a comment with your suggestions, or post a separate answer to the question.
If and when the team finally decides that making the rejection reasons for users' suggested edits more visible is a good idea, information like this will be very helpful.
If you feel particularly moved by a particular suggestion, and you don't think the person who suggested the edit will follow through after you reject it, you can add it as a comment or answer yourself. Nothing wrong with that, so long as you give proper attribution to the original user.
I've even done this occasionally when someone tries to leave an obvious comment to someone's answer using the suggested edit system. I will leave a courtesy comment to the poster, linking to the suggested edit, and summarizing their point.