Thanks for all the tips.
Further to your advice, I have connected all three GM harness grounds to one of the heads, with another dedicated wire from the same point all the way to the battery. I also ran another wire from the same point to the other head, so both heads are now grounded to each other and directly to the battery.
According to Chevrolet Performance, the MAP sensor should cause rough running, but not failure to start.
The main 4 AWG ground is from the block to the frame and from the same point on the frame to the negative terminal of the battery. About 12" for each cable.
The LiFePo battery has previously shown a lesser voltage drop than the "regular" Corvette battery and the Corvette battery is back in the Corvette, so using the LiFePo battery, the lowest transitory voltages with ignition leads disconnected (occurring immediately as the starter engages) measure as follows:
At battery, normal wiring: 7.x volts
Starter post to ground: 7.x v
At battery, w/jumper cables parallel to normal wiring: 8.x volts
Across starter posts when cranking: 0.03v
Battery jumped directly to starter, completely bypassing all car electrics: 10.8v, BUT the starter just spun at this point and did not turn the engine over. I assume it must have some kind of clutch or shear pin internally, which has just failed. This voltage result is interesting; I would have thought that with no load on the starter and the battery connected directly to the starter, the voltage drop should be negligible.
The above voltages are transitory bottoms; as the engine cranks, it immediately recovers to 10.x. It's a bit hard to read the transitory bottoms because they happen so fast and the numbers are not terribly repeatable within a 1 v range (so I have listed them as ".x")
Considering the voltage drops I am getting and the fact that it just broke, anyway, I have a new starter on order. Will report back once it is installed and testing can resume. In the meantime, I am googling Noid lights . . .