Ok, this is not addressing the fight issue which does certainly seem to be resource motivated and as such hopefully manageable if you control flashpoints. With my boys I know that I can feed my food guarder with only one of the others and that I can verbally manage Freddy if he moves in on Brice while Brice will concede food so no fight. The others wouldn't back down so we don't -*test*-('") it.
I have had a number of fights over the years (not just with the current crop of dogs) and they do seem to fall into "spat" which is sharp, short and quickly forgotten or "to the death" which is where I am at with Mugi and Brice - they are segregated 24/7.
A while back I was given a tip to separate fighting dogs and have once used a modified technique. I have also seen it in use once and it did work. Pet correctors, clashing pans and WW3 would not interupt a "to the death" fight BTW - nor will a hose, blankets etc. The main thing to remember is keep yourself safe too, however upsetting a fight.
So, the technique involves two slip leads. While the dogs are fighting loop one lead round the belly of one dog and drag the two dogs without trying to separate to a secure place where you can tie the lead (fence/tree etc) or if there are two people have one take hold and brace against something so they are firm! Then pop the other lead round the belly of the second dog. Then wait and watch while holding the lead loosely but firmly. At the first opportunity when the dog you are holding breaks for a breath (even a smidgeon) or the other dog looses hold fractionally pull the dog you have away, don't interact with it but just gradually drag it away from the opponent which is still tied to the fence or whatever. Try to put a physical barrier between the two dogs then pull yourself together before checking both dogs, once they are calm

for injuries. I have seen a single woman manage this with two powerful spaniel sized males who were very engaged in the fight, both dogs needed vet attention but the human was fine although obviously upset.