[porousmediumflow][2pncmin][nonisothermal] Use mass fractions for diffusive Neumann flux

In the non-isothermal 2pncmin test the mole fraction is used in the problem for calculating the diffusive Neumann flux of the evaporation rate. However, when using a mass-averaged reference system, the mass fraction should be used for the difference quotient of diffusive fluxes.

Therefore, I propose the following correction:

static const Scalar massFracRefH2O = getParam("FreeFlow.RefMassFracH2O");

...

const Scalar massFracH2OInside = volVars.massFraction(gasPhaseIdx, H2OIdx);

...

if (massFracH2OInside - massFracRefH2O > 0)

...

* (massFracH2OInside - massFracRefH2O)

...

* (massFracH2OInside - massFracRefH2O)

...

Moreover, the reference value for the free-flow in the input-file should be converted into the mass fraction.