Calculating SpecVol in a single location per timestep

[alicebarthel]

So far, individual developments that needs the Eos SpecVol have recalculated it (Steven's pressure gradient, Kat's vertmix).
Ultimately, we want a single SpecVol computation per timestep.

Options we have discussed with @cbegeman:

1. including the computeSpecVol in the timeStepper::init2(), then at the end of each time step (e.g. after/in tracerFinalize()). We'd need to make sure that the appropriate time level to be used.
2. including the computeSpecVol at the beginning of the timestep, but this would require implementing it in the different types of timestepping.
   Is there any expectation that we would need SpecVol to be calculated differently in the different timestepping schemes, e.g. any half-step calculation needed?

@hyungyukang has implemented a temporary functionAuxiliaryState::computeVertAux (here). This contains the necessary functions:

   Array2DReal LayerThickCell = State->getLayerThickness(ThickTimeLevel);
   Array2DReal NormalVelEdge = State->getNormalVelocity(VelTimeLevel);
   VCoord->computePressure(LayerThickCell, SurfacePressure);
...
   EosInstance->computeSpecVol(ConservTemp, AbsSalinity, PressureMidDbar);
   VCoord->computeZHeight(LayerThickCell, EosInstance->SpecVol);
   VCoord->computeGeopotential(TidalPotential, SelfAttractionLoading);

I just want to document this and open the discussion so that we make a concerted decision on the structure of our time-stepping. And that we go through the code and remove other instances of the Eos calculations to avoid wasting resources.