This issue can be closed with a set of existing methods applied in answering the following questions:
- Are two samples coming from the same disribution?
- How to detect stochastic dominance?
- Did the modality change?
- How many modes?
- Which quantile estimator to use?