How-to on handling large data

[nikhilk]

Creating a bug to track internal bug...

It's very difficult to understand anything about best practices for magics or modules without linking out to readthedocs.

This needs to be covered in tutorials and sample notebooks through markdown discussion and code and in help text for queries.
Specific worthwhile additions:

1. How to handle large data (common complaint) for in-memory work - when retrieved from BQ - this is needed for GA
2. How to handle large data in memory - Dataframe won't scale - this is post-GA and I have a separate tracking bug to use Graphlab's OSS alternative.

We should figure out how to better surface reference docs, as well as improve docs with a how-to set of notebooks to cover this sort of information.