Add feature distribution to partial dependence plots

It could be useful to add feature distribution (via histogram?) to our partial dependence plots so users can determine whether there is sufficient data to interpret the relationship between the feature and target.

https://doc.dataiku.com/dss/latest/machine-learning/supervised/results.html

https://towardsdatascience.com/prettifying-partial-density-plots-in-python-1f7216937ff