I don’t find the study ridiculous and I have met dozens of each breed you mentioned.
I find that most people assume the entire population of dogs grew up in a house and have been reasonably well looked after. In practice, I have seen lots of dogs that have that kind of history and lots of dogs that have had traumas that cause them to resource guard or be reactive to people and/or other animals (these behaviors are not bred, but are specific to individuals / life experience).
I totally agree that some behaviors that have been selected for have very high correlation with the breed, but lots of other behaviors are far less correlated. My Aussie will nudge me with his nose, more like the Corgis you mentioned and less like the Aussies you mentioned. Breed implies correlation of some phenotypes, not causation.
In the end, language matters. The title of the Nature article is a Rorschach. In practice, headlines are usually written by someone other than the author and almost always lose significant detail because they are short.
Definitions are really important here. I love herding breeds, but they were bred to nip at ankles and butts of livestock. Is that considered an “attack”? Does the person defining what behaviors are an “attack” have an incentive to exclude those small harm incidents? Chihuahuas are notorious for being aggressive and giving nips, but they present as small, quick incidents, whereas large breeds are more likely to bite down and/or swing the neck like a shark does. Does that difference bias the stats reporting/legislation choices?
Also, “Pit Bull” isn’t a breed. It is a vague intuition of some of the “bully breeds”. I always have to wonder if people pointing at attack frequency lists which show that inaccurate term are accurate. Did those lists gather data that actually investigated pedigrees or did they eyeball/guess breed based on assumptions?