WIP: [freeflow] Improve parallel velocities for stair-like boundaries
What this MR does / why does DuMux need it:
Fixes #888 (closed).
Special notes for your reviewer: Please only keep the second commit and omit all the temporary ones (MR set to WIP as long as temporary commits are in there). The temporary commits are meant to make review easier. In the following there is also the numbered grid for the kovasnay test which lets one compare the parallel indices.
Do you see any nice way to get an implementation for higher order without passing the curSol all the way down to staggeredupwindfluxvariables?
Edited by Timo Koch