Add pure bedload model

See merge request !138

The number of the shallowwater equations was hard coded to 3,
which caused a problem within the pure bedload model. Now the number
of shallowwater equations is got by ModelTraits::numSweEq().

This is a multigrain model, which does not include the
shallow water equations.

Add hiding and exposure

See merge request !137

The segmentation fault occured if the number of grain classes was smaller
than the maximum number of grain classes. The problem was, that in the spatial
params the vector bedloadformula_ is only initialized for the actuelly existing
grain classes, but in the volvars it was looped over this vector by the maximum
number of grain classes. This bug was introduced in commit 73668576.

The provision of output variables should not part of the layer model.
Instead it should be added in the problem.

improve secondary currents again

See merge request !136

This just consequent, since the bed shear stress is modified in the
presence of secondar currents. The velocity, which is also used within
the slope effect has already be modified.

Improve secondary currents

Closes #81

See merge request !135

The modified velocity is used for the slope effect as well as for the riemann solver

Up to now the bed shear stress was rotated by the secondary currents
angle, which is not correct, since the secondary currents add an
additional component to the bed shear stress.

The direction change due to the secondary currents is now applied within
the bedload transport, since we can use it here to modify the bed shear
stress, as it should be. Up to now it was applied to the bedlaod discharge,
what was not correct. Since the bed shear stress does not occure in the
Grass bedload transport formula we apply the direction change to the
velocity.

Improve assemble

See merge request !134

Fix some bugs in the initialisation of the hirano layer model

See merge request !133

Use an eps to ensure that layer 1 vanishes during the initialisation
of the Hirano layer model if it has the active layer thickness.

upperBoundaryLayer* -> upperLimitLayer*

Before it was 1e-6 m now it is 1e-3 m.

[limiting] Use LET limiter instead of linear limiting

See merge request !132

The LET limiter has more posibilities to influence the limiting.

Fix limiting bedload discharge flux

See merge request !131

The previous default values don't work for river models. Therefore
they werde removed.

The bounds for the limiting of the bedload discharge are dependent
of the massFraction of the corresonding grain class. This caused
negative sediment masses, since the bounds for the limiting became
small for small mass fractions. To avoid this we introduced minimal
bounds for the limiting, which depend on the representative grain
diameter.

Resolve "Use only the bottom shear stress of the friction law"

Closes #79

See merge request !130

This commit causes changes in the result, since the bottom shear
stress provided by the friction law differs since:
* Up to know only Manning with a spatially constant friction parameter was used
  Now the actuall bottom shear stress from the HN simulation is used.
* Also whenn using Mannings law the friction law calculates the bottom shear stress differently.

This avoids a division by zero, which causes NaN values.