Implementing OPE Estimators for Multiple Loggers

[usaito]

Overview

• Implement OPE estimators for the case with multiple logging policies, including Naive IPW, Balanced IPW, Weighted IPW, Naive DR, Weighted DR, and Balanced DR. These estimators are proposed in the reference papers.
• Add corresponding tests

Minor points

• rename obp.dataset.SyntheticBanditDatasetWithMultiLoggers to obp.dataset.SyntheticMultiLoggersBanditDataset
• some refactoring

References

• Aman Agarwal, Soumya Basu, Tobias Schnabel, and Thorsten Joachims.
  "Effective Evaluation using Logged Bandit Feedback from Multiple Loggers.", 2018.
  https://arxiv.org/abs/1703.06180

• Nathan Kallus, Yuta Saito, and Masatoshi Uehara.
  "Optimal Off-Policy Evaluation from Multiple Logging Policies.", 2021.
  https://arxiv.org/abs/2010.11002

[fullflu]

1st Review of estimators_multi.py

[fullflu]

I finished the entire review!

I also checked the whole lines of synthetic_multi.py (LGTM).