Amongst many other measures. I'm not sure how you can look at any graph like the ones shown and say that clearly masks don't work. I don't think there's really any scientific way of proving that poorly fitted masks work or don't work because of all the background variables which can't be controlled. If they do work, the likelihood is they have a minimal impact on transmission (but this could still be valuable), and the only way we would know is if we could compare an identical population in an identical environment with every other factor remaining the same which to me seems impossible.
I agree that correlation doesn't mean causation, for the reasons you mention and others, but the question isn't whether masks (and the other things) don't work, the question is whether they do.
If you're going to make something mandatory, and back it up with laws, police, fines and (in some cases, fortunately not here yet) prison sentences, you have to be able to show a significant positive effect for doing so. And the evidence is increasingly obvious that there isn't.