Below, I've linked the FABRIK papers, the first of which is particularly relevant to my question, as it shows the convergence proof for the single end effector case, and this makes sense to me. But I have trouble extending the proof and proving convergence for the multiple end effector case discussed in the paper. If someone can give a proof that this also indeed converges, that would be much appreciated!
http://andreasaristidou.com/publications/papers/Extending_FABRIK_with_Model_C%CE%BFnstraints.pdf