It's interresting data. But I think you are mistakenly reducing it to profit.
The caravan XP your companion which in turns allow them to solve issues for you.
Caravans raises scouting, riding and trading.
Positioning in a settlement raise charm, party role raise stewardships/medecine/scouting and engineering.
Leading a party raise leadership & tactics and roguery (aside from martial skills).
The competition between workshop & caravans do not exists : you do not share limited slots for both. You can have both and workshop are useless for developping companions.
The trade skill will only improves a proportion of the potential profit. But that potential profit is totally linked to what good is available and at what price, which is very dependent on which factions own which fiefs, and luck.
So I have to strongly disagree : caravans are not useless.
They turn capital into income that in turns allow you to stay in the green. They stock towns and they level trade,riding,scouting without taking neither a workshop slot, a party slot or a party role slot.
Finally, the survivability issue.
As it is designed, caravans survivability comes from it ability to dodge and outrun. It gets destroyed when it takes a wrong turn and get trapped. If you ever tried to play a raider playstyle attacking caravans, you know their speed is what allows them to get out of tricky situation versus a party that simpl is set to "attack" and doesn't try to out maneuver them.
More men will only turn "borderline" cases into "trapped cases" because they will always be weaker than full parties.
Even IF the caravan survive a fight, it now risks being burdened (they won't throw away their good, and they carry in proportion to their roster in town). And so it is now easier to catch, while not being stronger than what the current caravan are.
I believe that's why developpers said that it was the issue with your suggestion. It's a logical conclusion.
Your comparison of day 150 to day 750 forget one very important factor : bandits growth over time (as well as with player level).
For time only, it maxed out at day 900 (If I'm correct).
So the caravans surviving 87.5% over 150 days can probably be maintained by tuning the way bandits & looters forces grows overtime but that is tied to lots of other mechanics. If that's correct, modifying \modules\sandbox\ModuleData\partyTemplates.xml to get looters & bandits to have min=1 max=1 might show (by comparisons with the normal behavior) how much the survivability loss is impacted by bandits/looters growth compared to warring parties ?
It still will most probably be tuned and the caravan survivability will therefore be tuned as consequence. There is a similar issue with peasants parties being more and more often destroyed by bandits because of that issue. It comes from bandits parties of varying size "ganging up" : a small party runs fast enough to catch a group bigger than his, and the bigger, slower bandit group gets dragged into the fight, making it a win.
So the bigger the "range" of size for the bandits (I.E. instead of 4-> 10 it becomes 4 -> 50), the more those cases occurs. There are ideas how to go about solving that (my favourite : make it so that bandits "merge" parties. It remove their "speed" advantage, while making big bandits parties available for leveling lord's troops.)
That being said, if you find a modder willing to try :
LordConversationCampaignBehavior class, method conversation_magistrate_form_a_caravan_accept_on_consequence shows the process of upgrading a caravan into a "bigger" caravan, where the property "_selectedCaravanType" is set to 0 for a normal caravan and 1 for a bigger one. Adding code to manage a case where it is set to 2 and dialog to set it to 2 shouldn't be overly complicated rocket science.
it's good to see someone else trying to pin points complex issues about the simulation