The user can now provide a 'pipeline' function with only '**kwargs' in
arguments. It is more flexible and easier to write. The 'sweep_index'
keyword is always automatically added, so the 'passindex' option has
been removed since it was redundant with the index of the final
dataframe object.
The user-defined 'pipeline' function can now return anything. It is
no longer limited to ([x,],[y,]) format.