Well, I don't know why they chose to do that.
Some consequences of having a high ATV setting:
1. Servo resolution
2. Curve gets highly non-linear near the end-points due to rotary to
3. Increased transit times.
4. Some servo travel needs to be reserved for mixes.
Of course, non of these may be a problem if you know what you're doing and can work around them.
The 9Z is not designed to be free of limitations. It is full of them. Quite a few functions assume that you'll only ever want to hover at mid-stick. In airplane mode, if you are using the flapperon or aileron differential mixes, you can't use flap as a slave in a pmix - only one servo responds to the mix.
They could easily have made it much more flexible if they didn't make so many things linear developments from lower-end radios. The downside would be you would be in for a major programming session just to make a four-channel airplane work. The way they've done it, some things are very flexible, but others are pre-programmed and full of unecessary contraints. This acutally makes things more complicated. If you've ever dealt with assigning conditions to models with a Campac, you'll know what I mean. You can easily copy a model, but you really have to get into the guts of their memory allocation scheme if you want to customise your condition assignements. This is necessary as the simple copy model function is very wasteful of memory.
It seems like the 9Z was designed as they went along, rather than starting with a master plan for what they wanted it to do.
Well, you can probably tell that I've wanted to get that off my chest for a long time