Add AutoML Tutorials Around Best Practices

Add tutorials to: https://github.com/h2oai/h2o-tutorials that show best practices for AutoML.  Some ideas could be: 

* creating the correct validation data (not all situations should use cross validation)
* when to use weights column or sampling
* how to handle data with high cardinality categorical features or text