Possible bug in CSolver::SetRotatingFrame_GCL

[pcarruscag]

As mentioned in #975
This routine was averaging the solution at nodes i and j but it was updating the residuals at those nodes with the un-averaged solution.

SU2/SU2_CFD/src/solvers/CSolver.cpp

Line 2732 in 626be8f

LinSysRes(iPoint,iVar) += Flux * Solution_i[iVar];

Upon fixing it one of the turbomachinery cases started diverging, so I kept the suspicious implementation.

[economon]

Actually, on closer look, this seems ok. I am only averaging the grid velocities on either side of the face, not the solution. That routine might be more efficient as an edge loop, but should be correct. If you're really bored, see pg. 83 for details about the Geometric Conservation Law as implemented there: https://economon.github.io/docs/economon_dissertation.pdf

[pcarruscag]

I wasn't bored, but I'm always curious.

[pcarruscag]

Conclusion, it is not a bug, but there was a bit of code (the averaging) that was not needed.