It also depends on the author guidelines of each journal. Nature group e.g. has requirements about when your subscripts should be in italics. And i think sometimes they prettify images. I don't care about those things, and it doesnt change a thing about the impact of 99% of papers. Most authors submit great quality manuscripts anyway, judging from what i 've reviewed. In fact i m pretty sure i m doing more copyediting and suggestions while reviewing than any of the paid editors.
Another thing is, a ton of effort is wasted in reformatting (and sometimes rewriting) a paper to resubmit to another journal. Publishing is definitely a wasteful process.
There are always publishing costs, but it would be cheaper to set up a public utility for proofreading and uploading papers rather than the current situation
Take two scientific researcher of the same discipline & scientific skill level, and one of them might be a very poor communicator. Multiple levels of the editorial process addressed those issues, copy editing included.
Changes were not made without the author's approval: They were edited, proofs sent to the author, who signed off, or declined some changes while adding others. It would then be reviewed by the publisher's editors again, and if all went well a final proof was sent out for a final sign off.
I'm not sure why you doubt this sort of thing: I lived in that world, on the tech side of a small/medium scientific publisher, and saw it all happen & built some of the systems that facilitated the process. It wasn't wasted effort: margins weren't very big and where I worked they would have been happy to get rid of a few staffers if the process didn't require them.
If you published an article through a reasonably sized publisher in a somewhat well-known, did you different experience? If so, then either the practices vary greatly among different publishers, or practices have changed a lot since the early 00's (which may very well be the case)