It occurs to me that GF could do this in software, and create a perfect map of each GF. It’d be a pretty simple process - put in a full sized piece of material (paper, draft board). GB scores a grid (or at least a known set of calibration points) on the material. Then it looks using the camera, identifies the actual scored points, and does the math to correct for the optical distortion. Might want to repeat the process to confirm that the calibration worked. This should correct for whatever the unique lens positioning is in the GF - given the extreme angles, even a slight variation in manufacturing might produce a variation in optical distortion.
I remember doing this with scanners and cameras in “the old days” - print a calibration pattern on a known good printer (or pre-printed), scan it, and the software calibrates to correct for any distortion.
Amusingly cameras in orbit, or high altitude plans, use the highways in Kansas to measure optical distortion - the state is covered by a perfect grid!