Feser is actually wrong on a handful of points.

Science-qua-epistemic-method does not require "that this world is governed by causal regularities." In such a case, the method would still work, it would just figure out that all claims of causal regularity were false. Comprehensibility likewise shows his theistic bias: whether or not we can understand the rules of the universe are irrelevant to whether or not we can use science to reject our incorrect/limited interpretations.

What science does require is that the scientist not be actively deceived, (e.g. if our senses were telling us there is a world external to our mind but there wasn't, that would be a form of deception that would defeat science.)

How would you distinguish using science-qua-epistemic-method between a world where all claims of causal regularity were false and a world in which scientist were actively deceived?

The problem with the "actively deceived" scenario is that it defeats science, so demanding that we use science to determine whether or not scientists are actively deceived would mean that we have not quite understood the point.

Right - do you draw the distinction between the cases you propose philosophically then? I'm on board with an a priori commitment to the idea that we are not actively deceived in our observations of nature, but I think without the same a priori commitment to the idea of causal regularity in nature we'd have the same problem.

I suppose I don't see how "actively deceived" defeats science, but "universe not governed by causal regularities" doesn't? In both cases, we seem to be able to identify patterns and regularities in nature that are illusory.

Because as I said, we would do science and end up rejecting all claims requiring causal regularities. That we might observe something that seems like a causal regularity is irrelevant, because it wouldn't be repeatable, so we'd end up rejecting it. If you want to invent some case where "well, what if scientists measuring something makes it seem regular?" then you've created a causal regularity where we're not supposed to have any (i.e. scientists measuring causing regular behavior)

I'm also going to take issue with Feser using the concept of causation in the first place. While physicists certainly still use the language of cause and effect, I don't think they need the concept metaphysically in the way theists do.

Its like how biologists might say something like "this species of butterfly evolved a longer proboscis in order to drink from this kind of flower" but metaphysically they would understand that the butterfly species never actually had the intent to make their noses longer (neither did evolution, nor mother nature.) So biologists might use teleological terms as a useful shorthand, while understanding that there is no actual teleology metaphysically speaking.

I think physics is in the same place with respect to causality. They may speak that way, but at the end of the day they just need the data to fit their models. So we could have acausal regularities, and science would work completely fine.