[spekcular]

OK. The first paper is indeed literally about species and in a good but not top journal. But it's not connected to the mainstream of combinatorics in any way, in the sense that if you aren't a priori interested in species, there's no reason to be interested in the paper.

The second paper is interesting because it's in an excellent journal and about a problem not obviously connected with species. So I agree that it counts as a good example for your case. I also agree with your conclusion – it's an exception that proves the rule, so to speak. Most probabilists have no need for category theory, and most AoP papers don't use categories. (I feel that may literally be the only one?)

I also agree that, to the extent it's useful, it's useful as an organizing principle and not "substantively." I suppose this explains why I feel it's overhyped: mathematicians care about solving problems, and the insights that solve the problems ultimately have to come from some problem-specific observations.