* clang won't compile otherwise

* use Dune::Hybrid::elementAt() instead of std::get<>()

* due to a bug in clang (https://stackoverflow.com/a/47849501),
  myMultiTypeBlockvector.size() (which is static constexpr) has to be
  changed to MultiTypeBlockVector::size() to be used as a constexpr
  template argument
* probably a similar issue with using Dune::Hybrid::forEach and std::get<>

[fix][multidomain][stokesdarcy] fix convective coupling so that it is energy conservative

See merge request !1194

Feature/improve 2p1cni

See merge request !1191

* only build two executables

* allows to use thermal non-equilibrium model (T_soil != T_fluid)

* shall not be used but is required in order to fix clang compilation error

[io][gridmanager] Make gridData when making grid from dgf

See merge request !1190

[facet][box][fvgridgeom] use correct codim for facet mapper

See merge request !1186

[md][stokesDarcy] Correctly account for orientation of gravity vector

See merge request !1184

[staggered][freeflow] Print position of failing BC

See merge request !1185

[box] allow mixed-type boundary conditions for box

Closes #560

See merge request !1187

Test in test/geomechanics/elastic.

Feature/2p1c wettability

Closes #522

See merge request !1140

Feature/multiphase poroelasticity

See merge request !1157

[material][porodeformation] use correct law

See merge request !1183

[md][bugfix] Fix memory leak due to cyclic shared_ptr dependency couplingmanager<->problem

Closes #557

See merge request !1179

* hydrostatic pressure increases in direction of gravity

The 2p1c model only makes sense in a non-isothermal setting. The 
artificial isothermal base model is removed and numEQ, energyEQ,... are 
set directly now.

Dennis Gläser authored 6 years ago and Timo Koch committed 6 years ago

This adresses the memory leak associated with both the problem as well
as the coupling manager having shared pointers to each other and none of
them being destructed at the end of the simulation.