It's the static calibration method. The calibration method is like:
After a good six-sided calibration described above, the accuracy of the virtual sensor, Accelerometer Calibrated would becomes 3. Otherwise, you have to do the six-sided calibration again, because, maybe you did not put the device on a flat place or the calibration time for each side is too short.
I'm still trying to read out the calibration state from the BHI chip. See the "BHI260 - Calibration State - BSX Algorithm Parameters" thread (https://community.bosch-sensortec.com/t5/MEMS-sensors-forum/BHI260-Calibration-State-BSX-Algorithm-P....
Once I get that working I'll be able to validate the drifting issue.