Improve current state of tests

[pbrehmer]

Possible improvements:

• do a review of all tests
• get a better example model than the XXZ Hamiltonian (which is used for many tests, where a RBasis is needed); also the fact that the XXZ Hamiltonian has discrete magnetization plateaus and is very sensitive to the degenerate snapshot at $(\Delta, h/J) = (-1, 0)$ makes this model problematic at larger $L$
• write tests for untested code parts
• design an artificial parametrized eigenvalue problem, where we know the exact solution (non related to quantum spin systems)

[pbrehmer]

In addition:

• tests for resumed simulations
• consider the FullDiagonalization issues (see TODO)