Ensure correct solver scheme for PDA

[gassmoeller]

The entropy material model requires iterating over advection and Stokes equations (because temperature diffusion is computed split from entropy advection). This was not asserted so far and has let to problems with temperature overshoot in other solver schemes. Also assert that the projected density mass conservation formulation always uses a compressible material model.

@jdannberg, @RanpengLi does this make sense?

[jdannberg]

Conceptually, this makes sense to me, but sometimes we want to run a model and just check the initial conditions and therefore select the no Advection, no Stokes solver scheme, which technically is not consistent with any material model. For example, one might want to write a test that just checks the conversion from temperature to entropy and back. Or, someone might have a prescribed Stokes solution but still wants to solve the temperature equation using entropy.

Therefore, it would be good to either have an input parameter that does this check by default, but can be disabled ("Check solver scheme consistency" or something like that) or at least allow the no Advection, no Stokes option as well.

[gassmoeller]

Good point. What do you think about the current version? The only solver schemes that are not allowed are the ones that only solve advection equations once and move forward in time.

[jdannberg]

Yes, I think that makes sense!
This also leaves the option open to just set the number of Nonlinear iterations to 1 in an iterative scheme if someone really, really thinks that this is what they want.