LatentHeatMelt material model has a bug in thermal_expansion_coefficient

[bangerth]

Pointed out by @alarshi: This function doesn't make much sense:

    template <int dim>
    double
    LatentHeatMelt<dim>::
    thermal_expansion_coefficient (const double temperature,
                                   const double pressure,
                                   const std::vector<double> &composition,
                                   const Point<dim> &position) const
    {
      return thermal_alpha;

      const double melt_frac = melt_fraction(temperature, pressure, composition, position);
      return thermal_alpha * (1-melt_frac) + melt_thermal_alpha * melt_frac;
    }

The git log information is as follows:

b434d073 source/material_model/latent_heat_melt.cc (Juliane Dannberg  2014-06-19 18:40:51 +0200 143)     {
924f6d13 source/material_model/latent_heat_melt.cc (Rene Gassmoeller  2016-11-22 10:01:33 -0700 144)       return thermal_alpha;
b434d073 source/material_model/latent_heat_melt.cc (Juliane Dannberg  2014-06-19 18:40:51 +0200 145) 
b434d073 source/material_model/latent_heat_melt.cc (Juliane Dannberg  2014-06-19 18:40:51 +0200 146)       const double melt_frac = melt_fraction(temperature, pressure, composition, position);
b434d073 source/material_model/latent_heat_melt.cc (Juliane Dannberg  2014-06-19 18:40:51 +0200 147)       return thermal_alpha * (1-melt_frac) + melt_thermal_alpha * melt_frac;
b434d073 source/material_model/latent_heat_melt.cc (Juliane Dannberg  2014-06-19 18:40:51 +0200 148)     }

@jdannberg , @gassmoeller -- any insight into the reasoning why the first return was added?

[jdannberg]

This was fixed by #1706.