Licensing

[aazuspan]

Before release, we need to:

1. Select a license for sknnr.
2. Ensure we comply with the license of ported code. We currently port code from:
   1. yaImpute throughout, which is primarily GPL v3 but also contains some public domain code. Their GPL license was inherited and upgraded from the LGPL license of the ANN library.
   2. statsmodels for the implementation of CCorA (although we can hopefully migrate to the sklearn implementation of CCA), which is BSD-3 licensed.
   3. scikit-learn for the implementation of dataset loading, which is BSD-3 licensed.

See #27 for further discussion.

[grovduck]

For now, throughout the code, we have used the marker REFERENCE in docstrings to denote places where we will need to specify the licensing from ported code. As part of resolving this issue, we need to make sure that we remove these markers and replace them with meaningful licensing information.

[aazuspan]

Just came across this 3rd party license file and thought it might be a good approach to use for our licensing--basically have one Markdown file that lists all the licensed code and where/how it's used. We could probably add the full text licenses here too, as required.