I think in theory it would be fine to stop mid stream and wait for any amount of time and resume. I'm not sure that would ease your warping issues in all cases. What happens on long thin parts is that the the top layers shrink and tug at the layers below causing a slight curl or lifting force. Even if you stopped and waited for the raft or lower layers to stop thermal expansion and contraction, the upper layers are still likely to exert force as they continue to be thermally active. I've even heard of parts cracking the bed glass because the part was stuck so well that when the part began to shrink, it broke it.

Starting in the middle of a failed print sounds like something beneficial but that can be tricky if you don't mess around with code much. You'd need to find the layer and coordinate where the print failed and restart there. I guess you'd need to be careful to have the head come in a little high and then drop down to the start point so it didn't hit the part and knock it off of the glass. I don't see any reason that you can't do that if you do it with care and tweak the Gcode just right. It also depends on what the failure is obviously. Some failures cause such a mess on the top layers that you might have a hard time getting a plane that is all the same height upon which to continue printing. If you need to remove the glass and print from the printer to clean up that top layer so you can resume, you just need to find a way to register the glass on the heat bed so you can get it back into the exact same spot within a few tenths of a mm so that the restart matches where it left off.

I have not tested this but I have read a little about layer adhesion. One article suggested that after a layer cools beyond a point, that subsequent layer won't stick to it as well as as if it were still warm. I think some testing needs to be done on that. If you think about it, a layer gets pretty cool on a really large part before the head makes it back over there. So in that case, you are printing hot material on much cooler material. If the new layer does not actually melt the lower layer and bond with it what you get is a hot bed on top of a cool bead and as the hot bead cools, it is sliding along the one below it and never stick. As it shrinks, it is tearing away at any points that it stuck to, in effect, microscopically sliding along the layer below, thus creating poor adhesion. Your own tests will determine if you get layer separation. I'd be careful on parts that are under stress or are mission critical. It may not be a good idea to stop too long before continuing the print until you determine just what factors affect layer adhesion.

Unfortunately, each of us spends a lot of time coming up with techniques to minimize thermal expansion issues. I'd say go for it and try some experiments. If you come up with some good techniques please post them as we all be interested in them.