Airspeed calibration time seems annoyingly long

Does anyone else think the airspeed calibration time is way too long?

It seems the calibration is taking an average of 2400 samples. I realize many of the airspeed sensors used in the Pixhawk world are very noisy, but this seems like an arbitrarily large number of samples to average.

We’ve been testing a branch that cuts the number of samples in half (to 1200) and we see no noticeable difference in airspeed calibration quality. Would folks be supportive of an upstream pull request to shorten this calibration time?

how about a sample that depends on variance/noise? if readings are noisy, take more; readings are stable, take less?