Blurb::
Use a polynomial surrogate as the discrepancy model

Description::

For the purposes of model discrepancy, constant, linear, and quadratic 
polynomial surrogate models are available. These models are less expensive
to calculate than Gaussian process models but are consequently less flexible
to discrepancy trends that are not constant, linear, or quadratic in nature. 
For more information, see the Dakota User's and Theory Manuals.

By specifying this command, each discrepany function will take the form
of a polynomial regression. They will all have the same order.

