Very cool! I developed something similar for my Duo (separate thread over in the PowerBook forum), but although that platform has some of its own unique challenges (ID chips, current regulated charging algorithms, etc.), one critical factor made the single port charge / discharge steering easier than what you've overcome here: When plugged into the charger, the Duo's battery terminals see ~20vdc, which gets pulled down by a NiMh when inserted. That allowed for simple diode steering into a battery-internal Li-Ion 3S CC/CV charger (which requires a source voltage over ~13.5vdc to engage) when plugged into AC, but it isn't engaged by the battery alone when it's discharging and AC is unplugged. I have a functional 18650-based pack, but am not nearly brave enough to try selling them
I'm now in the process of developing a 2S pack with its own power-path steering circuity for my PowerBook 180c, and while in some ways its characteristics are more conducive (7.5vdc OE charged voltage is a great fit for a 2S Li-Ion's 7.4vdc), the single port charge / discharge switching is more complicated. I'm using a 2S CC/CV charge board (similar to the 3S part I used in the Duo pack), but with only 7.5vdc from the wall charger, a boost converter is necessary in this application (or a split as you've done, very creative!) I can't just let the Li-Ion cells / BMS back feed into the boost converter, and with a common charge / discharge port, some sort of sensing / switching is necessary to prevent that (as you're obviously well aware).
Ingenious approach with the split cells, comparator, and electro-mechanical relay. Love seeing resto-mod projects like this leveraging modern tech to augment these old 68Ks
In my case, I'm planning to use an LT1494 (or similar) current-sensing opamp to drive the gate on a FET powering the internal boost converter / 2S charger
only if current is flowing into to pack. Bench testing currently on-going with a thread to follow if it goes anywhere...