An excellent analysis. In my own modeling I have found that NS code 83 rail has about the same conductivity as #22 solid copper wire. My choice is to keep the rail sections short and to power each with a single feeder, preferably soldered somewhere near center. The longest is 18 feet, and the feeder also acts as a jumper across the third rail joiner, soldered to the outside of both rails.
My variant of the MZL control system uses common rail, but that rail is also gapped at short intervals, and each section has a single feeder. With individual feeders, each section of both control and common rails can be fitted .with an occupancy detector when signaling is desired - in my case, including five-lamp five-indication color lights with heads half a train length apart.
Here in the Dessicated Desert corrosion, even of half century old brass rail, just doesn't happen. I have a dozen yards of brass, laid at the bumper ends of several back-in staging tracks, that is still clean and conductive enough to keep the markers lit when I park brake vans on them. That area is cleaned, occasionally, with a Shop-Vac. (The rest of my old brass rail is scenic material, pretending to be the new CWR scheduled to move onto the concrete ties the first weekend of October, 1964. My signature explains why that makes the rail permanent scenery.)
Chuck (Modeling Central Japan in September, 1964 - TTTO, 24/30)