Decided to have a quick play this afternoon and see what's there.
The bad news is that the accelerator pedal position sensor is not a variable resistor - it's actually a sort of magnetic switch
out of the 6 wires that go in, it's actually 2 pairs of 3 - just as the factory wiring diagram suggests. I joined 6 wires into the sensor harness so I could experiment with the sensor plugged in and the engine running.
By directly shorting some wires I was able to get the engine to rev. I was working on only 1 of the pair of 3 wires. Despite only 1 of the pair being shorted, the engine revved quite a bit before the ECU noticed a problem and dropped the revs back down. I could do this 2 or 3 times before it engaged limp mode. This seems to suggest there is actually a bit of leeway in the sensor readings before the ECU detects a problem.
The "kinda" good news is that the sensor terminals themselves have a resistance value that is static. I should be able to make a basic PCB that would reflect those same resistance values which when connected would be treated like a pedal at the rest position. I would then need to change the resistance values slightly to hopefully get it working.
In case anyone else wants to have a play I've provided the values I found on the sensor below with the ECU loom unplugged. With the loom plugged in and the engine running there was 5v present at the wires indicated.