Perturbation order increases with iteration for oblique cluster shapes

[maierta]

For Nc=12, "cluster": [[2, 2], [-4, 2]], and likely other oblique clusters, the perturbation order is found to increase with the DCA iteration. This is likely because the chemical potential keeps being adjusted downward to get a fixed filling, i.e. it does not converge. I believe the cause of this is likely a problem with the new coarse-graining.