I'd stay away from any suggestions regarding modifying the stave assembly/glue-up. All you'd be doing is pushing the need for precision earlier in the process and increasing your production time (i.e.: if the staves are pre-contoured, then the glue-up step becomes a high-precision affair, where it wasn't before—you haven't gained anything in efficiency).
On the live-cutter front, have you considered using rotating drum sanders with very coarse grit? Sort of a surface-grinder approach for wood
You could take the entire height of the shell down at once (here I mean the length of the drum shell measured along its axis—I don't know the right term). Then you wouldn't need the "cutter" to move in the Z-axis with any precision... that movement would only be for positioning. You'd slide the sanding head into the shell so that it extends passed the bottom and top, then lock the Z in place. Then with he shell turning and the sanding drum turning, you only need to apply enough force in the X-axis to engage the grit's cutting action, and you could do the whole thing with depth-stops instead of any kind of precision feed. You may find that you don't actually need the shell rotation to be powered at all, as long as the rotational axis is fixed.
Might take some trial and error to find the right balance of cutting speed and grit to avoid tearout, but once dialed in, your only consumables would be the sanding drums.