Good point on the belt being cold making it particularly vulnerable in stuck-clutch-plate situation. I amended my comment to add the word "cold" to "That abrupt driveline shock is very hard on the entire driveline including a cold belt".
I agree that any and all snappy/abrupt increase in load is hard on belts and the examples you gave are good ones. (Though I'm really interested to hear in what terrain and riding style has you sitting still, redlining the bike, and dumping the clutch all day long).
I submit that hitting stumps, some clutch-dumps when getting unstuck, and occasionally not correctly timing the landing of jumps/drops to be "throttle off" are all contributing belt stressors, but those occur as part of a days riding. Shocking a cold belt and drivetrain, due to a stuck clutch plate is avoidable, so...avoid it. Ha. No reason to abuse it, even a bit, when it's easy to avoid.
Belts are very strong when under heavy load, but are not good at very abruptly going from slack to taught, particularly when cold as you mention. Abrupt loading events can snap belts, or can cumulatively stress them to the point they fail later, with much less force than their load strength indicates. Think of how hard it is to break a piece of monofilament fishing line when you slowly tension it vs giving it slack and then abruptly yanking it, which easily snaps it with a fraction of the "lb test' strength simply by decreasing the duration of the loading event. Spreading the loading event out over a few extra fractions of a second makes a huge difference, too. Dumping the clutch on a warmed up bike with warm oil and warm clutch plates seems immediate to us, but the amount of time the loading event takes place is not easily perceived.
It would be interesting to test and see data on how long the shock event/ belt loading takes in different situations; dumping the clutch on a warmed up bike when getting unstuck, clicking cold bike into gear with stuck clutch plates, hitting a stump/rock that abruptly slows track speed, landing jumps while still on the throttle into powder, same on-throttle landing onto high-traction snow, same landing but off-throttle, etc. And also testing shock strength vs load strength for both cold belts and warmed up belts. And if belt and drivetrain position makes a difference in the load event duration (meaning, the position of the belt/pulleys/sprocket/chain/drivers etc. after rolling a snowbike backwards out of a trailer vs after the bike had been driven forward first and then stopping which positions the tensions differently and would impact the load event).
Not going to have access to such data anytime soon, so can't come to a conclusion, but the point is, the duration of the loading event is a huge factor with belts and it would be interesting to see the impact of that as I suspect that matters even more than the work involved.
Anyway, I wasn't saying that stuck clutch plates were going to constantly snap peoples belts; only that it was an unnecessary, additional (read: cumulative) stress on the transmission and belt so avoid doing it.