Tensorial dielectric permittivity (refractive index)

[myurkin]

Is your feature request related to a problem? Please describe.

Simulating liquid crystals and magneto-optical crystals. It is also related to #12 (at least, liquid crystals with different director orientations).

Describe the solution you'd like

• Internally, consider m, epsilon, chi, and alpha as full tensors.
• Consider the implications for the performance (especially, in combination with #12). Either keep all of them always tensorial, or somehow switch between several options (like, scalar, diagonal, symmetric tensor, arbitrary tensor). Some of those has already been considered in branch wd.
• Specification from the command line may be reworked to avoid very long lines.
• Using non-symmetric tensors will break reciprocity. Not clear what are the implications. One thing is taking the sqrt(alpha) to symmetrize the system - this would not be possible (with certain consequences for iterative methods).

[TonyaManina]

I will work on it