Add a User Guide "How to handle large CSV?"

[zaleslaw]

Users often asks about limitations of KDF to handle large dataframes

The User Guide should contain some recommendations and snippets of code to improve User Path here

• some benchmarks on real-world or synthetic CSV files or datasets
• snippets of code to handle it with SQL databases or Apache Spark
• fine-tuning for KNB, IDEA or Jupyter Notebooks
• explanation of limitations of KDF reading model

Related to the #141

[Jolanrensen]

I tried reading the 800+ MB csv file from here and I keep running into OOM errors. It might be a good candidate for trying to get it into a DataFrame or working with it.

It contains about 34,959,672 rows of data.

(Edit: It only runs OOM from JUnit tests. It works fine with enough memory from a main() function or a notebook)