Interesting calculations. That makes sense. I would have expected the series connection to be more efficient. What is the installation? If they are not all pointing the same direction you may gain some efficiency using two separate MPPTs.
Due to having 3 of one type and one of another.
I would suggest calculating one of each type first, then putting result for the pair in series back in as a single panel + other pair to get 2s2p results.
As results as posted give strange config for paralel / series configs.