Why is this method considered a 'calibration' technique rather than a direct measurement of gyro error?

It's a calibration technique because it uses an external, known force (the sun's gravity) to induce a measurable error in the gyroscope. This error is then correlated to the gyro's performance. It doesn't directly measure the gyro's inherent error characteristics like bias or scale factor; it reveals how the gyro responds to external disturbances.