Before, it was a single number for the whole domain. It could vary
with time, though.
Simulator behaviour should be identical before and after this commit.
This touches the classes TransportModel*Polymer, Simulator*Polymer,
and the computeInjectedProduced() functions. The PolymerInflow class
and its usage has been replaced with PolymerInflowInterface, and
a subclass PolymerInflowBasic has been created which provides the
old behaviour (using parameters poly_start_days, poly_end_days and
poly_amount).
Note that due to a design constraint in the transport solver, namely
that inflowing fluid must consist of only water (possibly with polymer),
and no oil, the program fails unless the comment on lines 186-- is
heeded.
