Yes. In theory, a single point source is optimal.
As for passive xovers, they can in fact exibit some 'nasty' behavior in certain circumstances. Active xovers can be superior if done properly. Exceptional passive xovers are not inexpensive to produce either.
I'm not sure comparing a 'typical' full range to a 'proper' multi-way is really apples to apples. How about comparing a typical full range to a typical multi-driver or a 'proper' full range to a 'proper' multi-driver? While multi-driver systems do have some advantages in terms of feeding drivers the ranges that they are most linear and being able to contour the frequency response as part of the deal - there is always a tradeoff.
A properly designed full range system will always be in phase. A properly designed full range system can exhibit much better imaging (not always but in theory it can all else being equal).
Also, there is a limit to what you can do with multi-driver systems. Is a 3 way better than a 2 way? 4 better than 3? 5 better than 4? etc.? At what point are you monkeying around SO much that you've caused issues in the time domain, increased price drastically due to necessary xover complexity, etc? K.I.S.S. principle?
Like anything else, there are good and bad examples of both. Think of the best systems you've heard. From my personal experience, almost to a T they've been simpler systems, many line sources, full range panels, actively crossed systems with a sub and a full range driver like a lowther, a 2 way monitor with just a blocking cap and balance resistor on the tweeter, etc. The ones that did have multi-drivers were usually thing where the drivers were engineered to roll off at a certain point and the xover is only on the bottom end of the driver.
Why do 'monitors' have the ability to 'disappear' into the room? Part of it is small cabinet size. I firmly believe that part of it is also significantly simpler xovers.
When was the last time you heard a 4 or 5 way system that sounded as 'real' as some of the things I've described above?
/rant