Walberla: Initialization with viscosities and relaxation rates

[RudolfWeeber]

Allow initialization of Walberla LB with:

• Viscosity and magic number (defaults to 3/16). For eqn see LB specializations
• Or with the full choice of four relaxation rates omega_shear, bulk, odd even. See code for setting this in Walberla in the LB specializations.

The code is in src/walberla_bridge/LBWalberla(Fluctuating)MRT.hpp (actual LB specializations incl their constructors) src/walberla_bridge/lb_walberla_init.cpp/hpp (instance creation) src/core/grid_based_algorithms/lb_walberla_instance.cpp (Espresso side of instance creation and management) src/python/espressomd/lb.pyx/pxd: cython interface code

The rules for calculating the four relaxation times from shear viscosity and magic number are currently repeated in the LB specializations. This should be moved into a free function. It might make sense to introduce a type LBRelaxationRates with omega_bulk/shear/odd_even as data members. Alternatively, a std::tuple<double, double, double, double> can be used, but then the order is unclear.

[itischler]

I can have a look at this on coding day

[jngrad]

Implementing this ticket would require rewriting the LBwalberlaImpl constructor. Not too hard, but there is currently no use case for this feature. We will not work on this at the moment. If someone needs this feature, please let us know here or on the mailing list.