Enable sample trimming in DRLearner (#1021)

This set of changes enables sample trimming in DRLearner (and subclasses), following Crump et al. (2009). The user can specify a fixed propensity threshold below which to drop observations, or can have the threshold automatically selected using the rule from the paper, which minimizes the expected variance of the estimates.

Author: kbattocchi

doc/spec/references.rst

@@ -135,3 +135,8 @@ References
     Syrgkanis, V., Lei, V., Oprescu, M., Hei, M., Battocchi, K., Lewis, G. (2019)
     Machine Learning Estimation of Heterogeneous Treatment Effects with Instruments
     URL https://arxiv.org/abs/1905.10176
+
+.. [Crump2009]
+    Crump, R. K., Hotz, V. J., Imbens, G. W., & Mitnik, O. A. (2009).
+    Dealing with limited overlap in estimation of average treatment effects.
+    Biometrika, 96(1), 187-199.