At my last job we did some experiments with replicated surfaces using epoxy based materials. What we found was that we were able to produce replicated surfaces, using a large granite surface plate, with a flatness within .001" over about 3'. The biggest problem we had was the hardness of the surface. We were running an air bearing over the surface. If any grit got on it tended to get stuck in the epoxy and could end up scratching the way or worse the bearing.