[FEA] Split datasets in `datasets` package chronologically

[karlhigley]

🚀 Feature request

Split the datasets in datasets package chronologically instead of at random.

Motivation

Splitting at random is known to be a problematic splitting procedure, since it leaks future data into the past. A preferred approach is to split the dataset chronologically, so that the first 80% of the data is the train set and the last 20% is the test set.

[rnyak]

@karlhigley Yes, I think we can do that for the datasets with timestamp.

[EvenOldridge]

@bschifferer @radekosmulski This would be a good example. How to prepare data for RecSys.