I just got a review audit where the "late answer" was a duplicate of the accepted answer, but flagging it on that basis failed the review. That seems like a bug to me.
Question here: Is it better to call a function that doesn't have an effect at that point, IF it improves code clarity?
I found that by clicking the question to view the entire Q&A, and noticed that the copy was from a generic "user12345" account (probably with 1 rep as well) and the original from a real-looking account. Viewing the whole Q&A is often useful when reviewing, especially late answers (which are often repeats of earlier points, disagreements with those, or random addendums better submitted as edits). I looked at the duplicate and thought "weird, rep harvesting maybe?" so I clicked "needs moderator attention" and wrote "copy and paste of accepted answer". FAIL!
I think that this review type can't ever work, because to avoid the problem I found you'd need to make up whole new "late answers" which by definition the review bot can't know anything useful about. Either that or to pass you'd just need to take any action at all, including upvoting the original of the duplicated answer. Which I didn't do, BTW, because I had no idea or opinion about the question or answer.
(this is not a duplicate of "What are review tests", it predates that and focuses on a particular defect within the process. See the comments below).