Automatic clustering of density results.

[GoogleCodeExporter]

We are interested in discovering fixed points/attractors in the time series 
results, the first stage of this is the density graphs, second is finding where 
the time series cluster, and third is finding the transition probabilities 
between these clusters.

The second step is easy to do manually in most cases, but automatic discovery 
is preferable as it is quicker and less likely to introduce confirmation bias. 
Therefore we want to add a helper method for discovering these ranges. 

Original issue reported on code.google.com by [email protected] on 4 Apr 2011 at 8:43