Topic: Material orientation

Currently, the orthotropic model contains code for computing the transformations of the tensors, and asks the element for information.
This.. unfortunately won't work for me.

I want to rotate the tensor inside the layered-cross-section as well, so the strain vector must be rotated before it is sent to the cross-section, so that the transformations are applied in the correct order.



Material orientations also mean that we must stop accessing the material status directly, as it doesn't necessarily contain all the necessary rotations. So, LastEquilibratedNodalInternalForces is a problem. Easy solution is to just use InternalForcesVector
I realize LastEquilibratedNodalInternalForces can probably make a noticeable optimization in some cases (though it is typically insignificant) if the material routine is *very* computationally expensive. It is only used for computing the reaction forces.
But most of the work, setting up B matrix, etc. is performed twice anyway.
Instead we could for example assemble multiple vectors at the same time, for different numbering schemes. Or few known expensive material models could check to see if the new strain vector is the same as the previous one.