Unexpected imported scalar boundary conditions from nek

[stgeke]

In nekRS a boundary type has to be specified for each boundary ID (tag).
If we import the boundary conditions from nek, there are usually no IDs available (unless the mesh comes from exodus or gmsh). In this case the IDs are derived from velocity boundary type (v, W, ...). This implies that on a given boundary ID the scalar boundary types cannot be different. This limitation might be unexpected from a nek user perspective.

I cannot think of a good fix for this. At the very least we should check if the number of boundary IDs match and bail out otherwise.

Suggested workaround: define boundary IDs in nek and specify types in par

[fischer1]

THere is a simple work around that involves counting.
We should fix this to support Nek.

[fischer1]

We can do this on the Nek5k side. It's simply a matter of taking care of it prior to usrdat2