making column checking robust to passing pandas index

[alistairewj]

Quick fix - passing an index object to columns causes an error (e.g. t1 = tableone(df, columns=df.columns)). Added a check if it's a pandas index data type since it's probably very common to pass it.