My own experience suggests that it is possible to do speaker-level passive crossovers, but I must qualify that by saying that although I have put a lot into the crossovers themselves, and the wire,etc., etc., it has only been since I got the ML2s tuned into the system that I have gotten this level of reproduction with this type of crossover. Previously, this type of crossover has eaten dynamics to an unacceptable degree, and I went back and forth between throwing more power at it and just living with a better-sorted, range-restricted sound from "better" but less powerful amps. I have never thought to try passive controls "driving" passive crossovers, and nothing I have heard to date has made me want to try digitally-processed "crossovers", either. This is why I asked the follow up question of your other post about digital crossovers for multi-way horns, because I am still waiting for credible evidence that anyone gets music from such a set-up.
Every crossover has its own problems, however, and the active crossover, for all its potential, is actually more difficult to implement correctly, in my experience. I am not sure why you would want a passive pre-amp, but if you insisted, still you might use active crossovers and limit the passive part to attenuation.
I still think that the most potential in terms of "crossovers" lies in the most difficult route of all, namely the narrow band/dedicated amp route. There are two reasons why I have not gone this route myself: time and money.
Well, add a third reason: I got lucky, I guess, and I am getting very natural sound now, with my present "compromised" system. I guess I'll revisit all this when I am overwhelmed by the need for comparable performance below 40 Hz.