A simple anomaly detection algorithm that reports sudden changes in a time series. Takes as input a time series in the form of a double and a spike threshold. If any point differs from its neighbor by more than the threshold value, it is considered an anomaly. The array returned matches the input array on anomaly points, all other points are set to 0. Note that this only takes local information into account, so if the time series has, for instance, a high magnitude period in between low magnitude periods, only the start and end of the high magnitude period will be reported.