I agree that it's not realistic to stop development but delay is possible.
The easiest case would be some authority figure miscalculating the amount of uranium required for a bomb. Since uranium enrichment is relatively straightforward but needs a lot of power and space for the separation equipment it looks impractical, so people don't try. Would Manhattan project have been pursued if critical mass was believed to be 10x more than it really was? A side effect of this error is that more dangerous partial reactions would occur because people would store and handle larger masses without realising, and this may deter the survivors. Of course it may also show there's something wrong with accepted wisdom and prompt another look. This would happen anyway, and probably sooner rather than later because everbody knew that existing knowledge was incomplete.
OTL the German bomb was delayed partly by miscalculating critical mass, and partly by shortfall in resources, while the Japanese lacked the resources. Infighting and perceived lack of need also played a part.
Delivery systems also need to be developed and air superiority attained, or you'll need a sub on a suicide mission to get to the target port city. This could also make it look impractical enough that you spend your effort elsewhere.
So for me, delaying nukes is possible but not for long (my guess would by <5 years (started during the war but not ready in time) or maybe 20 years or so if not started until after the war (this assumes that the urgency to justify Manhatten project would be less post-war than OTL.