Ok look at it this way,
Power is limited by the ability of the engine to get rid of heat. Take any size engine and stick it in a poorly designed cowl and you will find that you can only sustain high power settings for so long before the temps hit red and power has to be reduced. So you could say that the cowl or cooling ability of the engine itself (fins, water cooled? ect) are what limits the sustained power an engine can deliver. Now take your big radial. Sure, it may put out more power than a "typical" cont or lyc, but it's cooling will be designed to take this into account. Assuming it is cowled properly it will run no hotter than a properly cowled cont/lyc so that aurgument is dead! sorry
Now take any aircooled engine and consider where most heat is generated ... the cylinders and heads. The case and oil system are designed to keep internal heat at a controlled level as best as possible and generally do ... but this is much harder for the aforementioned sticky out bits. Therefore we try to either stick them out in the airflow with deflector baffles or within a pressure cowl. On the ground with no airflow through the cowl or down through the fins, the whole engine will cool at a similar rate. OK so the cylinders/heads will be hotter, but they will cool at a "similar" rate (cylinders may cool slightly quicker I guess as they will have a greater surface area). However the cooling differential will be nowhere near as great as when they have a good airflow over them. It is this cooling differential that causes the problem. I suppose ( but don't know for sure ) that a/c like YAKs with closeable fins to reduce airflow through the cowls might have addressed this to a certain extent, but normally this doesn't apply. So, when you have really heated the engine in a high power climb (re glider tugging) and you suddenly reduce power and dive, the engine case will not cool anything like as quick as the exposed heads, and unfortunately, due often to poor cowl design (in many cases) nor will the heads. The cast iron cylinders between the two shed heat very quickly in the middle but the areas near the two ends more slowly due to proximity of ally case and head and also poorer airflow. Anybody who has ever had anything to do with cast iron wil know this uneven heat/heat loss will eventually cause a crack!
As I said before, In most "normal" operations
Say again Slowly may be correct that this is not a terrible problem. But in any extremes like tugging, para dropping ect it is very real. Also as I have said, it is something you should gaurd against whatever, so it pays to practice good proceedure anyway and be sympathetic to your engine ... be it in cooling, leaning, use of carb heat ... whatever!
IM