Computing smaller process tensors from larger ones

[gefux]

It could be useful to have a function that computes smaller process tensors (in MPO form) from larger ones by:

• "tracing out" late time steps
• "tracing out" early time steps (this could depend on the system Hamiltonian and Controls)
• coarse graining the time steps (for example: 3 timesteps with dt=0.02 become one time step with dt'=0.06)
• truncating the MPO bonds with relative and absolute cutoffs.

The function header would then look like this:

def reduce_process_tensor(pt, system=None, control=None, start_index=1, end_index=-1, coarse_graining=1, epsrel=None, chi=None):

This could be particularly useful if one has a library of different process tensors (as I proposed in #119). Then the library only needs to hold the most accurate, fine, and long process tensors. The coarser ones could then be straightforwardly generated from the finer ones.