Use gyro data for Turn Tilt

[aeraglyx]

Small refactor that makes Turn Tilt frequency independent by using gyroscope data directly. Functionally, not much should change and it gets rid of some variables. The numbers were chosen so it roughly matches the behavior at the default loop frequency. I only tested it inside at low speeds but the setpoint value seemed normal, so hopefully it's correct (I don't really use Turn Tilt).

One possible regression is that during high roll turns (like berms), yaw rate gets reduced (pitch becomes the new yaw). I suppose it could be combated by calculating the yaw rate as d->gyro[2] - sinf(VESC_IF->imu_get_roll()) * d->gyro[1] or from the VESC_IF->imu_get_gyro_derotated() function. But maybe it's not important and gyro[2] is enough?

For now I used an ugly ... / d->float_conf.hertz, I guess ideally we would have a variable for loop time in seconds that could replace these divisions?

Would something like this make sense?