sorry for the late reply.
I only tested the Z gyro because in our application, such jumps in the X and Y axis should not be a problem.
I can explain a bit about our application. It is a navigation application with IMU and RTK GPS.
When GPS is available, this kind of jump is no problem. However, the estimation accuracy of the gyro bias is lower because of the jump.
When GPS is lost, the IMU is combined with an odometer to estimate position. That is, the gyro is integrated for heading angle, and the heading angle is combined with the velocity from the odometer to calculate position. In this case, the Z gyro bias estimation accuracy and the stability of the Z gyro bias are critical.