Different scaling factor calculations for triangular elements

[jfussbro]

Currently we have two different ways of calculating the scaling factor for curvature in triangular elements.
For a 2D triangle we currently do this:

And for all 3D elements with triangular faces we do this:

The 2D way might be better, because it is independent of the orientation of the element.
Maybe we have to adapt the scaling factor computation for 3D elements with triangular faces.