Improved the specification of required input data sets which is
either a polydata or unstructured grid. Also fixed issue with
proper specification of target processes to aggregate to. Also
the output type is now fixed to be explicitly known so that
we don't have to specify the output type.