add variable distribution to partial dependence plots

[RoyalTS]

... as produced by model_profile either by marking off variable deciles at the bottom of each plot the way sklearn does it e.g. (https://scikit-learn.org/stable/auto_examples/inspection/plot_partial_dependence.html) or by including an entire histogram (example: https://towardsdatascience.com/prettifying-partial-density-plots-in-python-1f7216937ff).

[hbaniecki]

(Note to self) I would like to add this feature into dalex but the AggregatedProfiles explanation object has no information about the data attribute from Explainer. I am unsure whether we want to copy the dataset into an explanation object. A more effective approach would require e.g. computing histograms for all the variables during the model_profile initialization.

[pbiecek]

@hbaniecki is it feasible (possible with a new function that will get the data as additional argument)? if not then let's close this issue with suitable comment