Minor Description
I'd like to do some changes to FeatureField so that when the saturation function is used and the pivot should be computed automatically from index statistics, then things also work as expected with distributed term frequencies. It works by returning the feature term in extractTerms.
As a side-effect, no need to provide the IndexSearcher when creating the query anymore, it is autamatically picked up at rewrite time, which makes it nicer to use too.