[DOC] Best practices for model and score information access tutorial

[verakye]

**Is your feature request related to a problem?
The feature request is related to available documentation and tutorials. It would be nice to have a tutorial on how to best save and load models and scores, especially focusing on the fact that only the final model is stored in model of run_cross_validation but for return_estimator='all' the CV models are stored in scores. It doesn't become obvious from the current documentation that if the scores are then saved as e.g. a .csv file using pandas that one would loose the models from the folds.

Describe the solution you'd like
Clear documentation about best practices to handle this and where to find and access what information in the given back models and scores.