The fact that concensus can (and does) change is an important principle on Wikipedia. Any kind of double jeopardly rule would go against that, however restricted its application. I don't think it's worth it.