Implement the LocalOutlierFactor

I guess wrapping the sklearn functionality LocalOutlierFactor would be an easy addition to our outlier detection library?