I like the idea of measuring it on a bench with an electric motor and graphing the results. Very cool idea Mr. Fixit. You are a natural engineer! Some error from lack of loading on the drivetrain of course, but I don't think the gear contact surface friction (and distortion) is the biggest driver here. I would guess the only BIG variable left out of the bench test is the tire distortion, and that's not a huge effect on Cobras. Tires could lead to big losses through deformation if the tire loading is heavy enough. . . but I can't get much tire deflection on my Cobra with wide tires. . . not enough tire distortion to make a huge loss there.
No I believe the biggest loss is in fluid friction. It can absorb lots of energy trying to turn around all that thick gear
oil, hence the switch to lighter oils these days. That kind of loss, like the tire deflection loss, is definitely not linear, but could be measured on the bench. We are getting a pretty involved test though. . . much quicker to just load it on a dyno somewhere and measure it!