MCA convergence is using absolute tolerance rather than relative tolerance

From the paper:

However, as you can see [here](https://github.com/EnergySystemsModellingLab/MUSE_OS/blob/51f97ad485d2948b50a15282f9c8073dfa6d4f5d/src/muse/mca.py#L663) the model is actually using at the __absolute__ difference in the deviation between supply and demand (i.e. we're missing the division by Dk in equation A.33).