For full disclosure, I have a (not very popular) proposal up on Area 51 for Computer Science Educators templated after this site. I say this just to make sure any biases are clear from the start.
I, personally, think this is a thorny issue. Clearly theoretical computer science is math, and one could rather successfully argue that it's pure, absolute math in a way Physics and Engineering aren't. In that way, questions here on theoretical CS seem on topic.
However, as a matter of tradition, I should point out that there is Stack Exchange precedence for separating them. The Theoretical Computer Science Stack Exchange site exists, and is relatively active despite Math Overflow and Math SE existing. Even though most of those questions would, strictly, probably fit on either. In fact, there are certainly questions on Math SE about DFAs and Graph Theory and all sorts of great things that fit on CS Theory. So they're... on topic and not at the same time?
Another, much weaker, argument is that mathematics people and CS people frequently use different language. For any educator with an interest in going to a specialist CS/Math Education site it shouldn't be a huge deal to code switch. Still, I definitely find that both here and in my every day life math people and CS people can express the same thing in wildly different manners and having a place for people who "think" another way could be beneficial. Of course, a perfectly valid counterargument is that we want a wide range of expert perspectives here, and it's not one I disagree with!
My main concern is the sketchy Venn Diagram between the disciplines. There's an awkward scenario where you can find plenty of interesting, important questions that are off topic here ("What does the research say are good qualities in a first programming language for college students?" "How should I approach design patterns?") But plenty that are (such as the DFA/unreachable state question we had, which works on either). It seems to make sense to throw everything that can be here here, and then make the other site the remainder but it seems a bit sketchy to say "you can talk about teaching databases here, but relational algebra over there; this question goes here if it's about teaching how to write a FSM in Java, but there if it's about teaching DFA/NFA equivalence; concurrent patterns go here; concurrent algebra there" etc etc and it all gets a bit fuzzy and confusing.
I'm not sure there's a great answer. Overall, just like theoretical CS questions are still relatively on topic on Math SE, I vote that questions about teaching theoretical CS belong here. However, whether CS Educators is dead in the water or not (sadly, it looks dead in the water-ish to me as it stands), I do think it's at least a good idea to keep in mind concerns about heavily overlapping disciplines.