It has single precision support...

We run ut of PIDs during a typical run on KNL.

Additionally, uses the right number of subrefined intervals in the blockstructured case

This reverts commit b82e6531.

for consistency.

Otherwise debugging of nonlinear problems becomes hard.

otherwise my paraview throws me segfault.

To comply with latest pdelab.

- Use TensorProductElement in one example.

- We still use the sumfact option to swicth to the sum factorization
  code branch. This way it is still possible to easily swicth between
  sumfact and non sumfact code.

- Only TensorProductElements with the same degree in all directions
  will work. Anisotropie and adaption of quadrature rule will happen
  in the next commits.

So far, we used a global context to tell the normal visitor,
that it is generating for driver code. This was feasible, as
we only had one differing implementation. With more differences,
it is better style to derive from the visitor and implement
the differences there.