Technically speaking even 9A is going to be a killer. Why? From current boards supporting neopixles (DIYino, PL, NEC, Teensy), at least PL and DIYino support killing power to the stripes by the on-board drivers (programmable kill-key or stripoff, they are technically the same). Even advanced transistors in the size used for saber boards have min ~100mOhms of channel resistance, even if you put all 3 drivers in parallel, you are looking at ~30mOhms x 9A=270mV of min drop. That will lower your voltage right away from 3.7V to 3.43, which in turn reduces the current drawn, so there will be an equilibrium. If I factor in the IR drop over the wiring to the connectors and the pogos, then I can tell you for sure that even 9Amps is a gross overstatement, you will never reach that current (nor would it be healthy, such current over any of the current saber boards would instant kill it due to overheating, they are not designed for that kind of currents, nor would it make much sense). I can't tell you without making precise measurements how much the neopixel current limit will be, but my well informed gut feeling tells me it's not going to be more than 9A. More probably something like 6A.
As to saber boards supporting 5V with current regulation: never say never, but as stated many times, it does not make sense to supply neopixel stripes with 5V. 3.7V off from a single Li-Ion is just about the optimum for them. If you use a higher voltage, you do not gain brightness but waste energy. Not to mention that for such a board you might need a backback.
As a last statement: apart from all of this discussion, I love both your and TCSS's solution to pixel blade connectors. Never mind the Amps rating, the solutions with pogos and ring connectors are superb! Big thumbs up to all the guys who were involved in their development.