Tests of the resampling classes
Summary
This MR introduces the following changes
-
conservative tests for count resampling (Undersampling) with simulated data, that actually run it and ensure same result -
fix non-deterministic behaviour of current PDF resampling -
conservative tests for PDF resampling with simulated data, that actually run it and ensure same result -
count sampling and importance sampling with no replacements have a lot code copied from one to another. I propose adding a base classe that both of them inherit (hopefully PDF sampling will inherit it as well)
Relates to the following issues
Touches: #233 (closed) Touches: #152 (closed) Touches: #60 (closed)
Conformity
Edited by Ivan Oleksiyuk