Make as_table_columns works for other input types

Eventually, we want to allow users to call build_table on csv file. It would create a table object and predict the table columns for user.

While reviewing my work, validates that one can download an existing table into a csv file (includes RowID and RowVersion). He then create a new table with the exact same schema, and upload the csv (that contains the old RowId and RowVersion). A new table is created and the RowId and RowVersion was reset.
Example old csv:

Example new csv:

Python session:

supporting csv file via pandas: https://github.com/Sage-Bionetworks/synapsePythonClient/pull/546