I usually follow IANAP with HMB (hold my beer)
I actually was a physics student long ago but I had to drop out because I couldn't do the math nor afford grad school, so I became an engineer instead. Its easy to get wrapped around the axle when talking about differentials...

rather than add my own $0.02, I found this:
Open vs Locked Differential - Torque Transfer - Explained I think the kid in the video has got it essentially right.
This is exactly the video I needed to see. This answers everything. Wheel spin is transferred when brake is applied to free spinning wheel...that wheel spin is then applied to the ground in the amount it can (determined by a relationship between downforce of the vehicles weight and friction coefficient of the ground against the tire) thereby directing extra potential torque to the non spinning wheel. In lay person talk: torque is transferred across the axle when the brake is applied.
where I was wrong: this same thing happens with a locker. The locker doesn’t insist that it is 50% to each tire at all times because torque to the ground isn’t just a function of the driveshaft driving the differential...torque to the ground is a function of twist of the shaft PLUS downward force of the vehicle AND, most importantly, nonslipperiness of the ground.
Those two other variables...ground slipperiness (friction coefficient) and downward force/weight of the vehicle.. were the things I was missing from my mental model
All that said:
my lay persons understanding of what’s going on with crawl control, in all likelihood, is that the vehicle is braking all wheels and only letting small amounts of torque to each tire up to but not exceeding the amount of torque each wheel CAN transmit to the ground, given the dowsed force/weight on a tire and friction coefficient/slipperiness of the ground that the computer thinks the terrain will allow.
I.e. it gives each tire only enough juice that it can that will go forward while also keeping it from slipping. The moment it feels slip it’s says “ya nah,” and tries to give some juice to one of it’s buddies...likely also giving just enough juice to the slippery one to keep it at the same speed as the alternate tire it’s searching for to give torque to...that way you’re not dragging said slippery tire and wasting forward momentum or causing the vehicle to steer via stopped wheels as the computer blindly searches each wheel looking for the one that spins and also allows forward momentum.
what’s interesting is that whatever gyroscope that the vehicle has that is telling the vehicle that it is actually maintaining forward momentum...and not just spinning the wheels stuck in one place...must be quite advanced. It’s not like the car has eyes that it’s using to determine it’s physical location relative to the scenery.
also, this answers my question as to why crawl control seemed superior to triple and double locked vehicles in my videos I posted:
driver error. The locked drivers were exceeding the amount of torque that the slipperiness of the terrain and downward force/weight over the tires would allow...thereby causing their stickiest wheel to slip/spin without transmitting torque to the ground.
Crawl control, alternatively, was using the brakes to make sure that the stickiest tire DIDNT exceed its allowable torque, thereby allowing the vehicle to apply its potential torque to the ground without spinning a tire waywardly.
Had the triple locked Chevy driver applied less throttle, theoretically, he could have discovered the correct amount of torque that the terrain would allow and crawled up...
i suppose the point of interest there would be: what if crawl can vary the amount of torque at a quicker and more appropriate level than a human driver?
in situations such as that, crawl...or a vehicle with an automatically and partially engaging lockers combined with traction control...would be superior to the human driver with selectable lockers.
this would explain why Land Rover and others have integrated partially and automatically engaging differential locks rather than selectable lockers.