CSVs can be gross. sometimes they include rows with different column counts. I'd like a test that checks what Dataframe does when variadic row data is fed to the constructor, and a confirmation that our behaviour matches pandas.
This request-for-a-test grew out of #150
