Add a modified random forest estimator

[lionelkusch]

Keep a trace of the estimators after deleting it in PR #166

[jpaillard]

Following up on #166 I suggest including the examples_not_exhibited/plot_residuals_sampling.py, which shows how the modified RF can be used for conditional sampling. Method presented in Chamma_NeurIPS2023

[lionelkusch]

Note: The example examples_not_exhibited/plot_residuals_sampling.py need to be modfied because it use BBI which is not include in the library (see issue #42 )

[bthirion]

This is a perfect illustration of the fact that examples_not_exhibited is a terrible anti-pattern and should not stay in the mean/long term.