Functionality for large datasets.

 - Added several new options to handle data sets with many samples.
 - New options allow to specify to how many samples a dataset should
   be downsampled to, either with or without filtering. The default is
   using no filtering, which leads to much faster computing times.