ekf run output predictor immediately with new IMU data (#471)

* EKF collect_imu take const imu sample and populate buffer

* EKF calculateOutputStates cleanup

* EKF add calculate_quaternion output predictor method

* EKF: update documentation

* EKF: remove unnecessary getter function

* EKF calculateOutputStates only apply dt correction to bias

* EKF pytest assert attitude validity, not update() return

* EKF: correct documentation

* EKF: Do not make attitude validity dependent on yaw alignment status

Yaw alignment could fail in flight due to temporary loss of data and yet the quaternions would still usable for stabilisation even though the absolute earth yaw angle wrt true north was uncertain.

EKF/tests/pytest/test_basics.py

@@ -54,7 +54,7 @@ def test_filter_initialized(initialized_ekf):
     """Make sure the EKF updates after a few IMU, Mag and Baro updates
     """
     ekf, _ = initialized_ekf
-    assert ekf.update()
+    assert ekf.attitude_valid()
 
 
 @pytest.mark.benchmark