Add comparison of max idx and number of idx in model.fit

[monkey0head]

🚀 Feature Request

Check share of empty users/items history in train log. Compare number of unique idxes with maximal idx for users and items and print warning informing on share of "empty" users/items and negative effects.

Motivation

To make a preprocessing sequence:
DataPreparator (get spark dataframe with valid data types) - Splitters/Filters (split/crop dataframe) - Indexer (create indexers for train at first)