Updated README.md

[sourav-mazumder]

Added few extra comments/examples for setup.

1. Mentioned that R should be installed in the same machine where Zepplin process is running. This information is very important when someone wants to use this in a multi-node cluster
2. Mentioned that for any R package to be called from this R Interpreter the same has to be first installed using REPL shell of R.
3. Provided example on how to set Spark_Home as it is source of confusion sometimes.