Excellent points, Magfan. To further clarify my point about power, though, what I was basically saying is that for passive biamping voltage swing capability needs to be considered as well as power capability, particularly if there is a large disparity in the power capability of the two amps.
Using your extreme hypothetical example of a 10kHz crossover, the amount of power that would have to be provided by the high frequency amp is extremely small. However, if say a 50W amp were used for the highs and a 500W amp were used for the lows, the 500W amp would in effect be no more powerful than, for example, perhaps 75 or 100W. Asking it to provide more power than that would cause the OTHER amp to clip, resulting in lots of extraneous non-musical high frequency energy going into the tweeter.
Good point about latency, aka propagation delay. You're right, that never seems to be addressed. Not sure how it would play out quantitatively under typical circumstances.
Best regards,
-- Al