#247 Evaluate removal of spectral range estimate

[Ulf Lorenz]

The current function Operator::getSpectralBounds() has a few problems:
* While it is accurate for specific operators, it is generally too pessimistic as soon as you compose, e.g., a Hamiltonian out of a potential and some kinetic energy.
* It only makes sense for very few operators; basically, you only get useful results for operators that are either diagonal in the DVR (potentials) or the FBR (most (!) kinetic energies).
* For the Faber polynomial solver, we have alternative functionality that handles complex energies
* It is only needed for one special use-case (Chebychev solvers)

Consider dropping this functionality or merging it with the alternative functionality.