Active speaker, the future? or another fad?

Active speakers suffer a few marketing disadvantages. To name a few: expensive, difficult resell, don't allow playing with amps.
The advantages are obvious: 1) no need for mammoth amplification that's overengineered to reproduce dc to daylight; each amp can drive a specific frequency range 2) the crossover before the amp allows maximum efficiency and the amp has an easier load 3) cross-over behind the amp means that there are practically no shifts in cross-over frequencies b/ween drivers (passive x-overs shift with increasing power)... etc

Bi-amping/ multi-amping has been discussed a lot. That's the same principle as an "active speaker"

Exertfluffer -- all you need is the passive x-over's crossing frequencies and filter order/ type (i.e. 1st order Butterworth, 4th L/R, etc). You can take it from there with an active or passive line-level x-over.
It's useful to have some measuring equip though, or have access to same -- so as to check (at least) the amplitude response curves on & off axis.

Exert - the trick in bi(multi) amping is to know what the designer has done in the passive x-over. Not easy, I know. I wonder if Mr Thiel would give suggestions re, going active in the bass (if s/one were serious about this of course).

The point is, a line-level x-over is fed low level signals, and is much more stable in its filtering action (and often, cheaper). Also, an amp connected to the voice-coil (or nearly) is a much more efficient way of using amp power.

One of the problems with passive x-overs is they perform differently given the power fed to them (which is high), influencing the amps, the frequencies, creating magnetic fields, etc, etc, ad nauseam. If you get it right with a 1kHz sine wave at 1W, it won't necessarily sound "right" with music (multiple waves) at say 8W, etc.

Hence acoustic tests to fine-tune a spkr through the x-over

The bass is a very good first target for discrete amplification. There, all that's needed is usually the x-over frequency & slope; you'd also check for baffles step compensation and any other trick in the x-over (there won't be many). These points are easily addressed with an active filter (yr typical Behringer allows a choice of different slopes, frequncies of course, time delay, transform functions, frequency boost or attenuation etc). If the spkr manufacturer helps with info, it should be possible to get good results (probably better than the original passive) with more, but less expensive amps, by keeping the mid-high x-over as is passive, (one amp, low-medium wattage) and pulling out and biamping the bass region (higher wattage).

People like Nuddel (Genesis sp?) have been offering this, at a price!

why (...) NOT MAKE THE UPPER MID/TWEET DRIVER(S) ACTIVE AS WELL!!!?

Good point. As you imply, it's a compromise of sorts.
The reason is that most of the energy goes in the bass region, so the biggest immediate improvement would be an amp dedicated to that region. As less energy hits the mid-highs, distortion is somewhat less AND more importantly, a good amp can control the system without being a mammoth and devoting +50% of its power to the bass alone (i.e. so much lost for the mid-high region).
I.e., as you go up the spectrum there's less musical info, so your average tweet rarely gets hit with more than 5W, and that's playing the system LOUD.