As a former academic (grad student, not professor), and now professional software developer (so exposed to open source), the idea of self-plagiarism irritates me a little bit, because of its vagueness.
Of course, there is such a thing as "self-plagiarism". If you took an entire paper and resubmitted it to a different journal after it was published, that would be unethical. There is a related case that's fine where a book collects previously published papers, but that's done with full disclosure.
However, I also think that large portions of what is published should be reused, and the norms surrounding that reuse are suboptimal. The background section for two papers often should have many sentences or paragraphs that are identical down to the word.[0] I certainly remember some fruitless sessions wondering whether a vocabulary change between two papers was load-bearing or not--encouragine reuse and making it explicit when components were reused would make papers more readable.
In this particular case, I suspect that some of these reviewers were putting out shoddy reviews using copy-paste without thinking about it. But it's also possible some of them had just gotten good at copy-pasting the parts of reviews that are always boilerplate, and the non-repeated parts of their reviews are worthwhile.
[0] The fact that you often skip those sections when reading scientific or mathematical papers is an argument for reuse, not against it. In philosophy, those sections often aren't really skippable, because terminology too often has small idiosyncracies that are introduced early on.