It's not terribly difficult to prove that given any target there are exactly two trajectories from the shooter that intersect the target (first few pages of Cranz's Handbook of ballistics vol. 1). So, sitting in the bell tower there are two possible muzzle angles of the tank that would do the trick, one for each trajectory. Given the velocity of the bullet at the time it reaches the tank's muzzle, and the length of the tank's barrel I think we will see that the drop of the bullet (relative to the angle of the tank barrel) is such that a perfect shot is indeed "possible".

Reminds me of a joke...a Mathematician, a physicist and an engineer chaperone a camping trip. A brush fire starts in the night. The engineer reads the manual on the fire extinguisher and the physicist calculates the most likely direction the fire will spread. The mathematician wakes up, sees the engineer and physicist, and says "Aha! A solution exists" and goes back to bed.