replace by EnableGridGeometryCache

frictionlaw in spatialparams

to shearStress. Also in problem.hh computeBottomFrictionSource is
renamed to bottomFricitonSource.

For each element a object of a child class of FrictionLaw is
instantiated and stored within a vector.
This makes it possible to change the friction law on runtime, use also friction
laws which have more than one parameter, use spatial variying friction
values and even to use muliple friction laws.

The friction law compute now the bed shear stress instead of the source
term. So they return a quantity with a clear physical meaning, which can
be also used for other purpuses (e.g. sediment transport). Therefore the
computeSource method was renamed to computeShearStress.
computeShearStress uses now volvars instead of the explicit use of
waterDepth, u, v.

It is possible to extract the Scalar type from the NumEqVector type.
Therefore its not necessary any longer to pass it explicitly as an
template argument.

The source is now entirely calculated within the friction law.

One can run the test now also with Nikuradse. But be aware that the
friction value in the input file is a Manning values.
Addintional changes
- The names of the variables used for the calculation of the
  boundaryfluxes are changed, because they were not really clear. Also the
  documentation of their file is improved.
- Some smaller modification were also applyied, which were demanded
  by the discussion about the merge request.

Remove unnecessesary part from updateAnalyticalSolution.

Also some of the smaller suggestions of the review process are
integrated.

To improve the weird inheritance from Nikuradse to Manning a abstract
base class FrictionLaw as base for all friction laws is introduced, by
making computeUstarH virtual.

- use std::array instead of std::vector.
- In general manningN is a spatial parameter, so put it there, even
  if it is a constant in the rough channel test.
- Rename the water depth from h to waterDepth.
- Fix the dambreak test, which was incidentially broken by the work on thei
  rouch channel test.

The test provides a simple channel flow based on the friction law after
Manning. Normal flow is assumed, therefore the formular after Gaukler,
Manning and Strickler can be used to get an analytic solution.

Martin Utz authored May 09, 2019 and Timo Koch committed May 13, 2019

Add the entries for the shallow water model in module.txt.
Ensure that all classes and functions have docstrings.
Document the ShallowWaterFlux class.