Maybe it's because I'm a native English speaker with strong writing skills, but I don't think any actual editing of the text has taken place on the journal's side on any paper I've ever been on. Usually the co-authors do this in several rounds before submission.
It depends on the journal, but usually there will be subtle edits. In my experience, the edits usually make the article less readable, but they are common for major journals, at least in my field.
One of the major journals in my field has a ban on sentences that start with adverbs. I'll admit my writing often uses adverbs it shouldn't but the journal will silently turn sentences like:
> Typically, data are ...
into
> It is typical for data to be ...
instead of rewriting the sentence to avoid starting with "typically" in a more readable way.
At any rate, edits are pretty common, but they're usually to match internal style guidelines, as opposed to editing for readability/content/etc.
As a counter-example, though, I have had one case where a typesetter noticed that I had an incorrect paper in the reference list purely from the actual scientific context in which I was referencing it. It was a really nice catch, and interestingly, it implies that the person doing the typesetting has a _very_ deep background in the field and a very broad knowledge of the relevant literature. It's not something that could have been guessed from the titles of the articles I was referencing without knowing the subject matter intimately.
It's common for major physics journals to copy-edit before publication. The APS journals and the Review of Scientific Instruments both do it.
Sometimes they'll catch a subtle error, sometimes they make edits to bring the document into compliance with their own style guide, and occasionally they'll introduce an error. On balance, they make the publication much stronger and more consistent. I've gotten nice PDFs that highlight the diffs and any questions/concerns they may have.
This is a job nobody asked for but everybody has to pay for.
As an author I am very careful about what I say in my papers and how I say it. If something sounds stilted or odd, there's probably a reason. Words mean things and subtle changes can affect their interpretation. For this reason I do not like being edited by a non expert. If I feel my work needs such attention, I'll ask for it.
