This is a very interesting idea, but I believe the complications that arise make it less than useful in the end.
The first complication is the requirement to assume that the link will be correctly applied to the question. Answerers are not infallible, so we would have to deal with any questions that were given the link to no avail. There is no effective metric to assist in mechanically judging how related a particular link is to the question, as well; acceptance can change over time and votes may be more attracted towards a well-written answer than the answer with a link.
You also bring up the second complication, which is the fact that a particular link is not guaranteed to be addressing the same answer at all times. As a SharePoint user, I often see the same 2-3 people running about and posting the same software suggestions (typically custom written to apply to the question, of course). I also personally have linked the same tutorials for basic workflows when answering how to accomplish a variety of non-related tasks.
Perhaps the final complication is in the scenarios where this does work, it means that the duplicate has already received an answer. If it is a satisfactory answer, the author may be fine with the ensuing closing since the need has already been handled. But we now end up with the duplicated content we were trying to avoid in the first place (which also may make merging redundant). A post-resolution solution doesn't really help address the duplicate problem itself. If the author has already been satisfied (and by extension visitors with the same issue), then it is arguable how useful those linked questions will really be.
So with these 3 complications, I find that we would have a lot of noise in trying to use this constructively, and very little gain on any grounds.