I am generally not a fan of very steep slopes (eg. 48dB/octave) because (at least in the analog world of multiple feedback filters) the component tolerances tend to skew the results farther and farther away from the ideal. There are also possible problems with filter ringing (CCV filters) and under some filter topologies, instability.
I have found that in general, the point of diminishing returns is 24dB/octave. When you calculate power loss below the knee frequency, the squared factor works in your favor.
Interesting, so basically a cheap, analog filter resonates at extreme settings. This makes sense as to why this slope isn't commonly found in most standard analog crossovers. Do you find this less common in the digital designs which incorporate this as a standard setting? Do you find it a more prominent issue in the upper frequencies (rarely, if ever do you see this extreme a slope used in low pass or crossover points other than high pass). I'm wondering if that resonant decay from 48db to brick wall is why, with some systems, the extreme high pass results in an increased sense of "punch", or tightness, for lack of a better terminology. I also wonder what the actual decay rate of the resonance is, but am guessing that depends on the filter quality.