" Cell failures are rare compared with connection failures. " Hmmm...and you think there are going to be fewer connections with paralleled batteries than there would be with series cells? Only if you neglect the internal connections in that calculation. Good external conections are not hard to make, or ensure.
One also hopes that even with series cells, no prudent mariner is relying on ONE battery or bank. Even if their house bank is only one primary bank, surely there is a second starter battery which can be used to start the engine and provide power while a primary failure is being looked at.
“Regarding the loop between the two batteries "..."No, this is not tested, but is correct in theory."
I'd love to see you test that out, and see the results. Starting voltages, initial current flow between the two, and changes in that flow over time. Whether it goes down and stops, or goes down, levels, then reverses, etc. Since there is lag in each battery as the electrolyte locally saturates and then evens out again, the reaction will *not* be simple and uniform, each battery should be internally changing chemistry at different rates.