Pull out a one-liner for test scenarios to facilitate testing adjust()

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.

TargetTimeAdjuster will adjust a list of Features until it is within an optional
margin of a target total duration.

Helper functions:

 - _determine_margin() :: figure out the max and min cutoff times, considering
   margin and margin strategy (percent / absolute)
 - _features_total_time() :: basic sum of list of Features' durations

TODO: rename to TargetDurationAdjuster ? rename 'strategy' ??

Adjusters will be used to modify a list of Features. This could either be:

 - to modify the overall set (eg to target a time)
 - to modify individual Features

The most important Adjuster will be one that targets an overall time, eg:

"modify this list of Features such that their times add up to 1 minute (either ±
a % or a hard limit)"

@see: feature_extractors.py::FeatureExtractor