Potential outliers in summary output

[aalfons]

Suggestion by Patrick: When the number of observations is very large, it's possible that also the number of detected potential outliers is rather large. If that's the case, the output of summary() should only display a certain number of potential outliers (those with the lowest weight).

[aalfons]

Users should still be able to access the information of all potential outliers, so an accessor function potential_outliers() would be useful.