slow convergence for ccsdtq calculations

[binghuang2018]

Hi @Krzmbrzl @ak-ustutt

First of all, I have to say I'm really impressed by the efficiency of GeCCo (compared to mrcc from Kallay's group). And thanks for making this code public.

I've got one question regarding the slow convergence of ccsdtq for elongated C-C bond in c2h2 (R_CC=2.5Ang in my case), which is, of course, not unique to GeCCo, when starting from UHF wavefunction. I've tried to increase the DIIS cycles up to 20 (maxsub=20), but it doesn't help at all. Could you enlighten me on how to speed up convergence for this particular case? Thanks!