Adds support for R UDFs

[edgararuiz]

Builds the foundation that allows R functions to run in Spark/Databricks Connect

• It uses the rpy2 Python library to serialize and run the R code inside the cluster
• It uses applyInPandas() to run the R function (via rpy2) against a grouped data frame
• It uses mapInPandas() to run the R function (via rpy2) against non-grouped data frames
• Properly sets the max Arrow batch to partition de data
• Wraps functionality into spark_apply(). Most of the integration is in functions that spark_apply() calls, so the functionality can be re-used by future interfaces
• Adds rpy2 to the list of packages to be installed by install_databricks()/install_pyspark()
• Verifies if rpy2 is not install in your current Python environment and offers to install
• Adds tests