Tutorial on Implicit Q learning

Summary:
A simple example of how users can download offline data and train a Pearl agent with Implicit q learning, an offline learning RL algorithm.

This example uses offline data of transition tuples from a continuous control MuJoCo environment called HalfCheetah

Reviewed By: rodrigodesalvobraz

Differential Revision: D54936901

fbshipit-source-id: eff04dde2be8d42e650f7d5385ead27476d42dd0