my understanding is that string theory is basically dead, and only getting deader.

Huh... where is this impression from? String theory isn't dead, it's just a very narrow field in which most of the participants specialize in a subset of it that's less concerned with completing it as a whole. It's incredibly difficult work, progress is slow, and it's currently too broad to be applicable to reality (which is important for funding). The tests we can think of to validate the correspondence of math to the physical world are... significantly out of reach due to the energy requirements.

But it's still the leading theory of quantum gravity and there's active work in, say, AdS/CFT correspondence - which shows that string theory can line up to reality and be predictive. It's the best idea we have right now, it's satisfyingly elegant, and it's working as a useful tool at the very least.

There are competing alternatives that get their own research, of course. We should persue them all until a clear victor emerges!

But I thought modified gravity as explanations for the dark matter observations is seeing a bit of a resurgence lately.

Modified gravity, so far, is non-predictive and does not account for things like the bullet merger while also accounting for ultra diffuse galaxies and our observations of the CMBR. All proposed modified gravities have failed to pass experimentation compared to general relativity. Modified Newtonian Dynamics (MOND) fails in the face of light and gravity having the same speed. And even if MOND were to be true, it still requires the presence of (albeit possible baryonic) dark matter to be even considered due to existing mass measurements of galaxies.

So, again, dark matter is simply the best model we have.