The 1st stage is fine, full current until the pack gets to 3.5V x # of cells, but I think the 2nd stage should be to throttle the current back to about 5A, or so, until 3.7V x # of cells is reached, and then CV mode from that point (3rd stage?). Richard might need to chime in here, but I think that since the BMS board will start to take over, and throttle the current once the shunt thresholds are starting to be met, the Zivan should not terminate on the current dropping below a certain threshold (200mA, in your example...), but should probably just timeout after say, 3 hours. This should cover all but the most severely out-of-balance conditions, even for big Ah packs. If the charger times out before the BMS turns the green LED on full, the charger can just be recycled and it will get back to the same point pretty quick.
For the BMS control option, we could maybe use the same active low signal that we use to trip the SCR cutting off the charge current, to pull a line low at the charger. If it can read something like this, it can use it to stop the charge process. Otherwise, trying to do it just monitoring the current might be toughter. I suppose it could look for the current to go to zero for some period of time, like 5 seconds, or so.
I like the idea of having one 72V version, with selectable output settings, even with a 12A limit. Many people, myself included, have multiple setups with different pack voltages. BTW, this would work with 16-24 cells. Maybe they could do two versions, the 72V that can work from 48-72V, and a second one that could cover 24-48V, maybe with a bit higher current (15-18A?).