Add in anomaly detection

This release adds anomaly detection.

The Calibration object now has a new method raise_if_anomaly. This
raises an error if the magnetic field strength is greater than 3 standard
deviations form the mean (as recorded during calibration).

There is also a bug fix for ulabs np.cross function - this failed when the
passed data was a view of another matrix. This lead to incorrect reports
of the accuracy of data