I'm in two minds on this one.
Without splitting the shapes up onto layers , you can have no depth sorting.Also switching and shape transparency are a hassle.
It also takes longer to rig and onion skins are a problem.
Unfortunately the more layers you have the slower anime will run (true) and then there are issues with actions, shape interaction and losing compound shapes and drawing/construction, all which will slow you ( the animator ) down.
But then again it's easier to follow the point animation on say a head if it is on a separate layer.Then again,
again, the more layers there are the more faffing around there is to track down any point animation there is on each layer (unless you use group actions of course) for adjustment.
And you DO need to use point animation, this ain't thunderbirds.
There is no truly elegant solution to this yet IMO.
Just try and keep the layers to a minimum.
