Re: Why are physical checks needed?
"The physical test described would be far less accurate for any number of reasons "
The physical test would give REAL WORLD results, vs "optimal" ones.
Funnily enough, what actually counts in the end is..... REAL WORLD RESULTS.
Computer generated coverage maps based on topography are surprisingly accurate - even ones done as far back as the early 1980s. I drove surveys for a Telco when we were installing AMPS systems back in the dark ages and the holes that were predicted - and which we didn't believe - turned out to be 100% accurate (interestingly there were a number of dead spots in large flat areas of plains that weren't expected, due to long undulations the eye couldn't see but the mapping systems correctly marked as problematic)
Unfortunately those engineering-driven coverage maps got, shall we say, "heavily overoptimised" by marketing types to "remove" all those dead spots and make everything else look just peachy too. It didn't make them any less obvious to people driving down the highways in question, nor did it reduce caller blood pressure when helpdesks would tell complainers "according to our coverage maps everything is just peachy at XYZ spot, so there must be something wrong with your phone" (where engineering maps clearly showed no coverage - and that was with optimal antennas centrally mounted on the tops of vans and properly plugged into phones, not some handsfree thing trying to transmit out through a metallised windscreen.)