ScalableDataFrame should support convenient writers and readers: read_csv, sdf.to_csv, etc

## Proposed behavior

Using `Writer.of('csv').write(sdf, path='some_path.csv')` feels less natural than `sdf.to_csv('some_path.csv')`

However, current dataframe writers support much more: parallel reading/writing etc. These should be passed as **kwargs to the `to_csv` call.

This change also needs to be careful about inflight changes: #12 