All papers should be reviewed
That's a quite absolute statement. I can believe there are some situations where review may not be necessary. Of course, readers ought to know that and treat the content appropriately.
Of course, what this doesn't address is a paper getting quashed because reviewers refused to approve it,
Reviewers don't decide if a paper gets published. That's the journal editor's (or editorial board) job. A reviewer can put in an absolute stinker of a report, and the author not make significant changes, but if the editor (board) thinks it's suitable for publication, then it will be published.
I'm sure that happens from time to time. The question is how often? I suspect rarely. Nonetheless, I think that's the point of this change, so readers can see reviewers had strong opinions which the author and editor did not address.
possibly because the paper contradicted the reviewers' findings.
If that is the case, then the editor did an unprofessionally atrocious job of choosing reviewers.
I'll have to ask my kids (both in in the physical sciences) about how this works today. It's certainly the scuttlebutt I hear, that you can frequently guess who your reviewers are and that you won't make it past review if you don't cite the reviewers papers and you definitely won't pass if you contradict them. Yes, that sounds unprofessional and yet I can easily believe that's how the world works. People aren't as upstanding and unbiased as we want to believe.
(As you write, I'd like to believe this is more common in humanities than sciences. I have no idea whether that's the case.)
I can believe it because many fields are vey narrow. There may only be a dozen other people working in the same narrow specialty you're working in, which means there are only a dozen plausible reviewers. And given those other dozen researchers are competing for the same grants and publication slots, well, I can believe there's sausage being made.