Actually, shorting the terminals will do no damage at all, at least not to any Schmidt hub. And I doubt it would damage any other dynamo hub. It's nothing to worry about.
Also, the drag the OP noticed with the bike upside-down is real, but the effect is exaggerated if there's little or no mass being moved, as is the case with the bike upside-down. With the rider on the bike, the rider's mass acts as a buffer to the effect of the drag. Inside a SON 28 there are 26 magnets attached to the inside of the outer hub shell, between the spoke flanges. They are, in a sense, trying to keep the hub from rotating. As the hub does rotate, there are 26 points during each rotation of the wheel where the magnets are trying to keep the hub from rotating, and they alternate with another 26 points where the hub wants to rotate. As the hub rotates a bit in one direction or the other, the magnets are trying to bring the hub shell back to its earlier position. It's like riding down a road that goes up and down, up and down. Gravity tries to keep you in one of the low points on the road. So imagine that the road is essentially level over a long distance, but undulates up and down every 20 feet or so. From the top of each rise, you speed up just a bit for 20 feet until you get to the bottom, and then you slow down for 20 feet. This repeats over and over. The rider's mass is momentum, and the heavier the rider, the less he will speed up on those 20 foot down sections, and the less he will slow down on the 20 foot up sections. Over a long distance, there's hardly any difference in your average speed.
So, when you're riding a bike with a dynamo hub, with each revolution of the hub, you slow down and speed up 26 times, but since you have a great deal more mass at the rim rotating with you riding than with the bike upside-down, the effect, the amount that the wheel's speed actually changes 26 times per revolution, is tiny. Flip the bike upside-down and spin the wheel, you can easily see the speed changing.
In both scenarios, rider on the bike and rider off, the actual drag is the same. And with the lights off, it's equivalent to the added work you do climbing 1 foot of elevation every mile you ride. And it's about 5 times that with the lights on. Can you tell the difference if you're riding a road that's perfectly level, or if it's gaining 5 feet in elevation every mile you ride?
I didn't think so. ;-)