"but how does a tank open a door prior to going through it, short of firing a shell?"
Indeed, how do they do it? When a tank "drives through" a door, it doesn't leave the door flattened or wrecked. It doesn't literally drive through it. It opens it. Somehow.
So what's the difference between that usual scenario, and one where there's something standing on the other side of the door? None from the realism standpoint. The only difference is an interface one: one requires RMB, the other requires left-clicking past the door. And there's nothing natural or necessary about it, since in vanilla X-com you could open a blocked door just fine, using the same mechanism as usual: try to drive through it. So again, just an interface difference: openxcom doesn't let you do that if you know an enemy is there.
But
if you didn't know that, if the enemy was in the same spot but undetected, openxcom would let you! Now stop and think about that for a moment. The physical situation is exactly the same. The tank can't and won't actually move, any more than with the enemy detected. But the door opens. How? And since there's a way, what's stopping the tank that knows about the enemy from doing whatever-it-did-before-that-caused-the-door-to-open? The only difference is knowledge, not the physical actions available to the tank. Does knowing there's an enemy at the other side remove the tank's hands?
(And yes, I know the reason is pathfinding)