Should we necessarily close a question on meta if it's a duplicate of a question asked years ago?
No, not necessarily. If something's actually changed in the intervening period of time that would suggest that the conversation would somehow be fundamentally different if posed now than it was originally, it stands to reason that a separate, independent question might be a good choice in some cases.
However, this is often not the case for bug reports like the one that you've used as an example. While it's true that there are occasionally site changes that give cause to review previously dismissed bugs, in most situations the circumstances surrounding the issue were the same then as they are now.
Though it's possible that the original report was forgotten, I'm inclined to believe that it's really just remained a non-issue (otherwise, there would have been consequences). You could argue that if that were the case the original question might be sporting the infamous status-bydesign tag, but I don't feel that that's strictly necessary.
Granted, the two top answers on the original question do make light of the situation by using similar techniques to the one reported to give "joke" replies. But even those replies, along with the others, still highlight two important points:
- The problem can manifest itself in different ways, each of which would need their own fix
- There's nothing to suggest the issue is harmful to the point that it's worth the investment to fix
Based on that, I don't see any reason not to close as a duplicate given that this new post doesn't really bring - to borrow from what Won't said - anything new to the table, and I apply a similar reasoning in the more general case.