Scoring model as a good practice

[ArturoAmorQ]

Though in almost all of the notebooks we score the presented model somehow, there are some exceptions:

• logistic_regression.py (not really important as we have an intuition by eye but would be a nice-to-have, or least we can add a Warning message, I think)
• linear_models_ex_05.py and its solution
• trees_hyperparameters.py

On the same line of thought, ensemble_hyperparameters.py computes cv scores for parameter tuning but it doesn't pass the best parameters to a final test-set scoring.

Plotting a validation curve using the train set but no further scoring on the test-set (as in Solution for Exercise M6.03 and Solution for Exercise M6.04) is discussed in this forum post. In this case we can either use the whole data-set or add a Warning message.

We also don't score most of the notebooks that use the full data-set for modeling (see #441), but that is perfectly fine.