I started the close voting on this question, this was done under the needs details or clarity basis.
Another user commented they were struggling to reproduce your issue, and asked if you could provide a minimal example, this comment seems a clear and reasonable request for more details which was not acted on.
I have had no issues with references to two-author publications in revtex4-2 so upvoted that comment and, as three years seems ample time to provide more details, voted to close.
The question simply does need more details, just a single bib entry would have helped rule some things out, while a compilable example would have led rapidly to package versions which was apparently the real issue.
Closing a question isn't permanent, a closed question that's edited will automatically go into the reopen review queue and normally be reopened if improved.
It does seem (from additional information you are providing now) that there is a reproducible problem with a clear answer, that may lead to its being reopened as is.
Given it's been reasonably closed as unclear though, I'm not convinced it's worth reopening just to post notification of a bug in a version of revtex that was on CTAN for a week four years ago.
While such an answer would have been great while the question was open, I see nothing in the accepted nor the most upvoted answer to Questions which are bug reports, revisited to support reopening an unclear, closed question to post such an answer.
Your comment did not give any kind of solution, it just claimed to have tracked down the issue, nothing in it improved the question or evidenced that you had a genuine answer.
When another user commented asking if you had solved the problem, you did not respond.
It looked instead like you had abandoned the question, and that any future user would have been better off asking a new question, with more detail, as the close reason would hopefully encourage them to do.