Add wrapper around qdyn_optimize with krotov's optimize_pulses interface

[goerz]

Similarly to #9, we could implement a routine optimize_pulses that mirrors the krotov package's optimize_pulses function. Instead of using the Python implementation of Krotov's method provided by the krotov package, it would generate a config file and runfolder (exactly as in #9), and then call qdyn_optimize on this data before reading in the result from disk and returning a Result object. Obviously, there would be some limitations, like no support for non-standard info_hook, modify_params_after_iter, and check_convergence arguments.

I would advertise this approach as the recommended way to do optimal control with QDYN for "standard" problems. At this time, it would not cover "molecular" systems (anything with a spatial grid), although I could imagine how one would extend qutip.Qobj for such operators as well, which would allow a qutip/qdynpylib-based workflow for such systems as well.

I already have an implementation of this function in my own project codes, but it has to be generalized slightly before it is ready to be added to qdynpylib.