How can we get the size of a dataframe in PySpark