[test][freeflow] Improve error calculation.

Change of the calculation of the L2 error: The calculation of the L2 error before built the four vectors staggeredVolume, errorVelocity, velocityReference and directionIndex. There where two loops: one filling those vectors, one using them. The reason for this approach was that for the staggeredVolume the two staggeredHalfVolumes corresponding to two different scvfs have to be added.
This MR changes the error calculation to one loop making use of the fact that both staggeredHalfVolumes are multiplied by the same factor errorVelocity[i]. So building the four vectors staggeredVolume, errorVelocity, velocityReference and directionIndex is avoided.

Adding a calculation of an L infinity error

Move printError out of individual test files to one common file.

Change error output format to a CSV file
ToDos
Manually make sure L2 errors are unchanged on a sample basis:

Angeli 
Channel 1d
Deal with following files:

test/freeflow/navierstokes/convergence.sh 
test/freeflow/navierstokes/channel/pipe/convergencetest.py 
test/freeflow/navierstokes/sincos/convergencetest.py