Check for compatibility with sklearn functionals.

[timmens]

Change API to sklearn standard --even though treatment effect estimation is a mix between supervised and unsupervised learning and therefore any estimator can never utilize the features of sklearn it eases readability and consistency.

To do:

1. Construct estimator class cforest which implements a fit and predict method.
2. Hide all other functionalities
3. Check if there is any compatibility with sklearn.

[timmens]

In commit (d95a7cfa0cc019b10bd0738fbd186f4e73c81613) I implemented a wrapper class CausalForest which implements the methods fit, predict, save and load.

What remains to be done:

1. Check if I can hide private functions
2. Check for compatibility with sklearn