More flexible coupling for new science?

Can we generalise the coupling  to allow new sorts of experiments. E.g.

 * Apply a scaling or offset to selected fields, perhaps geographically varying. Read a file with scale and offset?
 * Replace a selected field (or region) by data (pacemaker experiments). 
 * Could we have both a model ocean and a data ocean passing different fields to the atmosphere?
 
See also #12