fix: corrected continuous discrete update

[iandareid]

Fixed the continuous update according to @bsutherland333's analysis and retuned the estimator.

[bsutherland333]

That is technically correct, but your naming is a bit confusing. I personally would multiply the IMU process noise by the IMU timestep when you construct $Q_g$, meaning $Q_g = \sigma^2 T_s$, and then use $T_s (Q + G Q_g G^\top)$ for the propagation equation, where $T_s = T_{s, imu} / N$.