Indeed. There’s really no purpose in trying to figure out how much roofspace there is in the US. You only need to know what the average height of a building is. The roof is roughly the same as the horizontal space, and the amount of energy used in the US is going to be linked almost directly to floor space, so it’s just a ratio of roof space to building height that we care about. If your average roof can only supply X% of the power for an average building, then that’s a decently rough estimate of your answer.
This is certainly true, but you can at least get a maximum value. And more importantly, the fact that nearly no buildings do actually have solar panels on the roof is essentially all the evidence you need that it’s not really cost effective. Expanding the coverage will bring in more juice, but per unit you’re getting the same effect as one building with one covered roof and that’s been demonstrated to be non-worthwhile.