Causallib

Changes:
* New models:
* Matching (estimator and preprocessing transformer)
* Overlap Weights
* HEMM
* Weight models now have same `fit()` API as outcome models
* Updated dependency
* Dropped seaborn
* pandas at 0.25
* scikit-learn at 0.25
* Additional fixes and maintenance

Changes:
* `datasets` module with toy datasets for causal analysis
* NHEFS data from Hernan & Robins' book
* Simulation benchmark data from the ACIC 2016 data challenge
* `contrib` module for new state-of-the-art outside contributions
* [Adversarial Balancing](https://arxiv.org/abs/1810.07406) model
* New implementation for MarginalOutcomeEstimator (formerly UncorrectedEstimator) using WeightEstimator API
* Additional Jupyter Notebook examples
* NHEFS (Healthcare data)
* Lalonde (Economic data)
* Additional bug fix and documentation