Causal-curve

-------------------------------
- Overhaul of the TMLE tool to make it dramatically more accurate and user-friendly.
- Improved TMLE example documentation
- Much like with `scikit-learn`, there are now separate model classes used for predicting binary or continuous outcomes
- Updating documentation to reflect API changes
- Added more tests
- Linted with `pylint` (added `.pylintrc` file)

-------------
- Fixed bug that prevented `causal-curve` modules from being shown in Sphinx documentation
- Augmented tests to capture more error states and improve code coverage

-------------
- Removed working test file

-------------
- Added new `predict`, `predict_interval`, and `predict_log_odds` methods to GPS tool
- Slight updates to doc to reflect new features

-------------
- When using GPS tool with a treatment with negative values, only the normal GLM family can be picked
- Added 'sphinx_rtd_theme' to dependency list in `.travis.yml` and `install.rst`
- core.py base class now has __version__ attribute

-------------
- Added support for binary outcomes in GPS tool
- Small changes to repo README