Simplify code using skrub TableReport and TableVectorizer

- Add a notebook + video to show how all the pandas code in the [Visual inspection of data subsection](https://inria.github.io/scikit-learn-mooc/python_scripts/01_tabular_data_exploration.html#visual-inspection-of-the-data) can be simplified using [skrub.TableReport](https://skrub-data.org/stable/reference/generated/skrub.TableReport.html):
- Replace `ColumnTransformer` with [skrub.TableVectorizer](https://skrub-data.org/stable/reference/generated/skrub.TableVectorizer.html) starting from the [Using numerical and categorical variables together notebook](https://inria.github.io/scikit-learn-mooc/python_scripts/03_categorical_pipeline_column_transformer.html)
- In the same notebook, section [Fitting a more powerful model](https://inria.github.io/scikit-learn-mooc/python_scripts/03_categorical_pipeline_column_transformer.html#fitting-a-more-powerful-model), replace `OrdinalEncoder` by `skrub.ToCategorical`.
- Explicitly mention that `TableVectorizer` makes the column selection automatically by using its `dtype`
- Introduce concept of "low/high cardinality" and demonstrate effect of `cardinality_threshold` on the "native-country" column in the Adult Census dataset.
- Update visualizing scikit-learn pipelines video to use `TableVectorizer` (with scikit-learn version >= 1.8)
- Modify wrap-up quizzes that use the Ames Housing dataset i.e. M1, M4 and M5 to select subset of numerical columns with pandas
- Redo the datasets description using `TableReport`