This question already has an answer here:
When an edit is suggested to one of your posts you get a notification asking you to review it which gives you the option to reject the review. The problem is that many suggested edits get approved extremely quickly so that by the time you get to review the edit it has already been approved. Of course you can still roll it back, but that doesn't revoke the +2 rep that the user gained.
I'm aware of the feature request posted by BoltClock
But my feature request is specific to the case where you are rolling back an edit to your own post, in this case the concerns about revenge rollbacks etc. obviously do not apply.
As an example I was recently received a notification:
In this case someone noticed that a link I posted in my anwser was broken (or moved behind registration wall), and just replaced it with another link (not totally random, but also not correct either). Besides the standard problem with robo-approvers, to a causal reviewer it is hard to notice that it wasn't a valid link, which is an additional reason why I think that the authors rollback should act as a final veto for the review.
As I mentioned earlier I am aware of BoltClock's feature request, however while implementing that feature would make this one unnecessary, so long as it hasn't been implemented this request is not a duplicate.