Restructure sparsecoding/data

[belsten]

Right now, sparsecoding/data contains dictionary loading code, dictionary files, and transform code. This makes it less intuitive to use as code to transform data and load specific types of data are under the same umbrella.

I propose consolidating code such that there are sparsecoding/data/dictionaries.py which manages the loading of dictionaries (what is currently in sparsecoding/data/utils.py which can also use documentation), and sparsecoding/data/datasets.py which manages the load of datasets (contains the code in sparsecoding/data/datasets/bars.py and sparsecoding/data/datasets/field.py.

A sparsecoding/transforms directory can then be made that contains the all of the transform .pys. Additionally

This should probably be done after @HarrisonSantiago merges his contributions to the whitening code. Also, we'll have to update the example notebooks to reflect these new changes.

[belsten]

closed with #76