Add instructions to configure IPython/PySpark for python 3, now supported with Spark 1.4

[donnemartin]

Reported by core_dumpd on Reddit /r/DataScience.

Solution seems to be discussed in Stack Overflow here.

core_dumpd reports the following works, need to confirm and update repo:

I end up running this:
PYSPARK_DRIVER_PYTHON_OPTS="notebook --profile=pyspark" /usr/local/spark/bin/pyspark

With:
PYSPARK_PYTHON=/opt/anaconda/bin/ipython PYSPARK_DRIVER_PYTHON=/opt/anaconda/bin/ipython

I'm running on docker based on sequenceiq/hadoop-docker:latest with Spark/MiniConda added on top. The only real config options in the profile are for the ip = '*' and open_browser = False.

[donnemartin]

Commit 1405875 adds links to Stack Overflow and Reddit which provides instructions