Outline for the main encoding example

[LilianBoulard]

As discussed with Gaël, here is a proposed outline for the main encoding example. Please submit your suggestions!

• Start with a dirty dataset, probably employee_salaries
• First, use the TableVectorizer, just to say "with little trouble, it works ootb"
• Then go into what it did, and the internals (GapEncoder, DatetimeEncoder)
• Use the MinHashEncoder for a larger dataset (maybe the largest one we have)

Notes:

• Should we talk about the SimilarityEncoder?
• Should we introduce the problem of dirty categories in this notebook?
• We should integrate the cross-validation example from Examples overhaul #546, but not sure where

cc @myungkim930

[LilianBoulard]

Moved over to issue #662.