Add reminder about validation curves' dependency on generalization performance metric

[ArturoAmorQ]

In the Overfit-generalization-underfit notebook we present the validation curve of a DecisionTreeRegressor() using the Mean absolute error to score the model.
This could be a good opportunity to remind people that scores and errors cover different ranges of values and, therefore, training and testing curves can swap their relative positions depending on the evaluation method. I think this could partially help clarifying the notion of good vs. bad fit.