I only noticed a handful of screenshots that you posted, were there more elsewhere? I don't think that is a relevant sample size to base the fix off of.
I'll continue to look into this, but there are just too many cases of values not being in the 'absolute value' order.
http://everquest.allakhazam.com/db/npc.html?id=6472
http://everquest.allakhazam.com/db/npc.html?id=6092
http://everquest.allakhazam.com/db/npc.html?id=5866
http://everquest.allakhazam.com/db/npc.html?id=8539
http://everquest.allakhazam.com/db/npc.html?id=5398
http://everquest.allakhazam.com/db/npc.html?id=28381
http://everquest.allakhazam.com/db/npc.html?id=6681
http://everquest.allakhazam.com/db/npc.html?id=32799
http://everquest.allakhazam.com/db/npc.html?id=5384
http://everquest.allakhazam.com/db/npc.html?id=5390
These prove the 'sub-sort alphabetically if a tie occurs' is not correct either:
http://everquest.allakhazam.com/db/npc.html?id=5385
http://everquest.allakhazam.com/db/npc.html?id=5383
I am doubtful on Velious being a special case, but I will try and confirm that later by looking through Kunark / Classic content. Granted I'll be using the 'live values' shown on ZAM as a guide, but if any exceptions to the rules are verified by classic screenshots, that should show that absolute value sorting is not reliable.
I'm feeling a bit more confident in sorting by 'Faction ID', but I wasn't verifying that above with each link.
Part of me wonders if there was more beyond listing 'primary faction first', they certainly seem to list factions they are allied with 2nd most often (such as Coldain being primary, then Dain showing 2nd hit. Same if Kromrif is primary, Kromzek would often be next). Maybe it lists all factions the mobs will assist then other factions? Not sure if mobs assisting multiple factions is classic or not.
Will continue to look into this, I welcome any feedback or noted exceptions that break any proposed theory.