Added s3 + spark session instructions

[ying-w]

Instructions seemed outdated so I've updated it to use spark session.

I couldn't find a way to set hadoopConf from sparksession

Setting aws-java-sdk in --packages shouldn't be necessary since dependency resolution of hadoop-aws should grab the proper version, however, the version required is quite old (march 2014). I tried a newer jar but got an error that looked like this

[parente]

LGTM though I don't have a way to verify the steps. Let's merge it and fix over time if people find problems. 🍰

[ogierpaul]

Well, thank you so much @ying-w for adding this comment:

 # !ls /usr/local/spark/jars/hadoop* # to figure out what version of hadoop

That just saved my life.
I was trying to connect to AWS with jupyter/pyspark-notebook, and everywhere the code found was reference Hadoop 2.7.3. Turned out that my version of Hadoop in the latest docker image was 3.2.0.

Thank you again!