devinplombier - we did explore low level crossovers. Net result is that we knew our niche which was all analogue and purist. Other folks have done versions of what you suggest. We always looked at overall cost effectiveness and separate amps, cables, etc. are not cost effective solutions, in our opinion. Also, the time of the CS5 development was 1988 - early and crude digital performance.
Of additional interest may be that the level of control needed for net resultant 6dB/octave slopes over approximately 7 octaves is far from trivial, even with digital modeling. We committed to minimum phase x time aligned performance with the 1978 model 03 and never looked back. Note that one would need to model, in addition to the driver roll-off slopes, the various reactive circuits to cancel resonances as well as the impedance correction shunts to maintain Thiel's resistive (non-reactive) load profiles. Most folks are very surprised how difficult that is with off-the shelf digital filters.
The approach we were intrigued by is low-level, active analog circuits before the power amps. In fact one of our first, pre model 01, trial products in 1975 was just such a speaker with 3 built to order amps and custom active crossovers in a 10" 3-way in a large bookshelf format. That product was unfeasible for a new company in a farmhouse. The technical aspects were manageable and indeed prototypes were impressive. But we lacked the stuff for market education and penetration. Indeed, powered speakers never became very popular.
To your question: we envisioned a lot. Line sources, spherical globes, di and bi poles. Active equalization was our first market entry and contained enough uniqueness to create more demand than we could meet for the next 30+ years.
Cheers, TT