If the third detector was supervised the same as the first 2 and the fire power was supervised separately on its own zone then wouldnt it meet code? Its just a slight tweak to the wiring diagram. UL defines a circuit as a pair of wires with the power being the pair supervised by the EOL Relay and I believe that meets NFPA 72.
It is the same for an SLC circuit where you have additional power for say a duct detector or Convertional Zone Module and you supervise the power separately then the detector which is supervised via the SLC Controller board.
I'd have to get 72 out to make a specific citation, but the trouble condition does not annunciate for the affected circuit/device. The schematic posted generates a trouble condition indicating the circuit with the EOL relay installed having the issue, even though it could be anywhere else.
A SLC with it's associated hardware and installation methods is an entirely different beast.
With that many smokes, if absolute necessity to know what triggered the alarm, given the amount of hardware involved, I'd consider a small 25 point addressable panel for fire alarm. Barring that, I'd wire as 2 wire circuits with logical zones/groups of detectors, then if using System Sensor I3's, I'd use a 2w-MOD2, a RRS-MOD and then, instead of using 16 zones, figuring this is a normal 2 story house, at most, you'd have say, 4-5 zones of detectors? Using the devices listed, then I'd wire a zone for the maintenance signal.
IMHO, it simplifies the wiring, uses less aux power to run, uses less components to properly wire and supervise, and give you a few other perks/benefits...and in my case, factoring in 16 smokes, additional power supply/battery sizing concerns, and 16 EOL relays....it adds up to be a cheaper solution going the route using the hardware I threw out there.