But as I see it, those forces are internal and should cancel out.
If we imagine the car going to the right.
When you brake the caliper generate a force going down. This could be seen as a force down, plus a torque in the wheel.
The caliper gets the same force up. Which could be seen as a force down and a torque in the upright.
The force going down to the wheel transfers to the bearings, to the hub and it's cancelled with the one for the caliper.
Still, as you said, I have that torque in the hub which I wouldn't if the brakes where inboard. That torque would be clockwise as seen in the image.
Now we have a similar situation with inboard brakes:
The car is going to the right.
As brakes are applied you get a force pointing down on the disc, and one pointing up in the caliper.
The force at the disc can be imagined as a torque and a force on the center of the disc.
This force would cancel up with one in the caliper giving a net moment at the chassis, halfway between the calipper and the center of the disc. Is this right?
Anyway, supposing what I said is correct, I still can't get why with inboard brakes the brake force is applied at the wheel centerline.