Am I interpreting 6150/9150 gyro bias specs correctly?

I see in 6050/6150/9150 product specification PDFs that initial ZRO tolerance is +/- 20 deg/s and it says ZRO can vary 20 deg/s over temperature range. Questions:
1) Are these gyro bias specs? I assume "yes" and they aren't some MotionSensor SW spec.
2) gyro bias:
a) I assume this is a 1 sigma spec so it can be larger
b) when my production software calibrates the part, I need to handle up to maybe 25 deg/s gyro bias (since it is just 1 sigma) in my software and not call values less than that "broken part".
c) Also my Kalman algorithm would need to be able to converge, even with very large gyro bias.
3) gyro bias over temperature: I calculate that 20 deg/s over temp range -40 to +85 gives .16 deg/s/C. This seems very large compared to your ITG part which says .03 deg/s/C. So is it .16 or .03?