Still think it's mostly a radiated interference problem. The fact that it all works lying on the bench
but not when assembled tightly into a flight box impliesthe interference generated by the camera only becomes significant when very close to the GPS receiver.
So put the camera in a seperate faraday-shielded enclosure to reduce radiated interference as I suggested, and use ferrite beads on the leads to it to reduce conducted interference.
I suddest you only leave the lens unsheilded, but even there you could put earthed wire mesh over it. The focus of the lens will be at infinity, and a bit of mesh over it will not appear on the image - try it.
regards
Allan..
ps I spent years in the RF business, and solving EMI problems like this was often necessary.