My opinion is that the other specs of the survey ship - speed, range, endurance - matter more than the time it spends loitering in orbit. Doubling the sensors (and keeping all the other parameters the same) would require doubling the rest of the ship - double the engines, double the fuel, double the crew quarters, etc. So, one sensor per surveyor.
I disagree with your math -- or maybe just the way you wrote it. I agree that travel time, deployment length, etc., all have a huge effect on how fast one ship can survey a system, but that doesn't mean that simply sticking one more sensor on it (with no other change) can't have a noticable impact.
Conveniently, Aurora tells us how many points are required to survey a grav location or system body, so we can do some quick math and figure out how many hours a ship spent sitting still surveying (the only place additional sensors help). Watching fuel can tell you how many hours were spent moving around. The ratio should give you a decent indication of where to best improve your survey efforts.
I'd like to get a ball-park estimate to know whether or not my Survey Ships are over-kill or under-kill.
With a conventional start, I build a 'single geo sensor, single small engine, absolute minimum everything else I can get away with' ship as quickly as possible to survey my home system. My real survey ships take years longer and tend to have two of everything (because my empires believe in a little redundancy, and because my fiction imagines a certain amount symmetry in my vessels).
Virtually evey survey ship I ever design for any empire/game gets a variant designed with "just one more sensor" or "one more engine" or "one more fuel tank" etc. I keep checking the effectiveness these theoretical designs to see if I should modify my real survey ships.
For example, if adding a third geo survey sensor (with no other changes) drops the speed of the ship 2%, it's probably worth it. Whereas if adding another standard fuel tank boosts the ship's range from three months to four at full throttle, but drops the speed to the point where life support won't last that long, it's a bad idea.
- - -
The fastest first-order-approximation method is to compare fuel used to life support duration after one system survey. If your ship burned 1/10th of its fuel and half of its deployment time, you probably want to triple (or more) its sensors.