One observation: you are wrapping too much chain around the crank. i.e. the test is not as close to “real life usage” as it could be. Why not wrap the chain from 6 o’clock to 12 o’clock as normal(course, depending on BB drop vs rear axle height and rear cog size)?
In fact, if I explain this well, you might improve readings by using standard chain wrap. My thought is that by wrapping the chain 75% around the big ring, you could be over-bracing the big ring and reducing meaningful readings. Why? If your measurement point is 6 o’clock then you’re looking for flex about the axle in the vertical plan. But to me the extra chain wrap from 12 to 3 o’clock is bracing the crank in this same vertical plane. Hope that makes sense.
Also, why not measure the crank arm deflection? To me, your measurement point is only looking at spider and BB flex. Why not put a gauge right under the pedal hole of the crank? That would measure straight vertical flex.
While you’re at it, why not add 2 gauges pointing at the side of the crankarm by the pedal hole lying horizontally? This would give a concept of the crank arm twisting as it was loaded up.