Returning true propensity scores

[MaximilianFranz]

Usually, treatment is assigned based on a probability by sampling from a bernoulli distribution or a binomial distribution with n=1 draw. Thus, instead of returning the actual treatment assignment, or additionally to it, the propensity scores / probabilities of treatment can be returned. Thus, the methods could, for sake of evaluation, use the true propensity score.

This might be a desirable feature for academia, while the practical applicability is questionable as propensity scores are rarely known.