Pressure acceleration for BoundaryModelDummyParticles

[LasNikas]

As discussed in #223, it is not clear which formulation for the approximation of the pressure gradient has to be used here

TrixiParticles.jl/src/schemes/boundary/dummy_particles/dummy_particles.jl

Line 106 in 341b921

@inline function pressure_acceleration(pressure_correction, m_b, particle, particle_system,

There are either

1. Symmetric formula: p_a/rho_a^2 + p_b/rho_b^2
2. density-weighted pressure: p_ab = (rho_b*p_a + rho_a*p_b)/(rho_a + rho_b)
3. or (p_a + p_b) / (rho_a * rho_b)

The 1st one has to be used for SummationDensity and the 3rd one for ContinuityDensity, right?
The second one is the one which Adami uses for his boundary model and is also used for edac scheme and the edac-boundary pressure acceleration.

My suggestion is to dispatch pressure_acceleration on the different pressure gradient formulations and not for the boundary model at all. That is, we don't have to distinguish if the neighbor system is a boundary system or not. But this could be a problem with the BoundaryModelMonaghanKajtar, right? Do we find a solution here?

[svchb]

Yes the same formulation has to be used in the boundary and rhs.
If we are at it as discussed in #132 we should also at the formulation average pressure version. Anyway I only wanted to add the mathematical correct version for SummationDensity and ContinuityDensity. We should probably move these forms than to a common place. But the current version if Erik fixes them in his PR as I discussed with him this should be mathematically correct.

[efaulhaber]

All these questions have been answered in #132. Also see the corresponding tests.