I cannot say how they are done other than walking around, but at a guess they are less required these days, mimo wireless has led to beam steering, which means you can get a lot furthur away and still be able to connect.
I would caution that rf propegation through 3d space is hard, almost every case i have seen had the measuring setup changing the results when rotated. The gradients still exist, but at certain locations reflections catch up, signals that where delayed by different densities in concrete interfere, etc
Using a spectrum anaylyser however would help find all the competing noise sources much easier.