Sorry for not approving this one earlier, I missed it somehow.
Long story short, climate used to only have 2 samples per year so the resolution of the itcz was pretty low. Now we take 12, which helps, but we still need to rebalance model parameters for the higher resolution to capitalize on that increased capacity.
Generally, your diagnosis of whats going wrong is right on the money, though.
I'll be fixing it sometime soon-ish