Tries to drop the lowest-scoring Features until the target time (range) is
reached. This is not optimised and a relatively naïve approach- there are many
inputs which would result in a non-ideal pruning.