I dismantled the motor. There was an obvious bad wire where it exited the axle as I was getting an intermittent 'judder' of the motor before it cut out, which was affected by the angle of the cable. So I shortened the 9 wire cable making sure to cut upstream of the poor connection. Not sure if this was a short or a broken wire.
This didn't fix the issue but was clearly something which needed to be addressed.
I then tested each hall sensor. I don't know if maybe I was turning the wheel too far when I tested previously but it was really clear that two were fine and one was not - the yellow wire. I tested continuity on the pcb from the soldered connection to the sensor and the yellow had no continuity while the others did. The middle sensor is fitted in the opposite orientation to the outside two. This seemed to be a clear break in the track on the pcb, so I soldered a jumper between the yellow wire and the sensor as in the photo. The motor then worked!!!
I don't think this was caused by a poor connection in the 9 wire cable where it entered the hub. I suspect the issue which caused the hall sensor anomaly was a poor track on the pcb which was intermittent for a while and eventually split as the bike was ridden over rough ground.
Anyway, problem solved and I can use my bike again!!