There is no technical reason why the end user couldn't map out corners and turns in a CAD program. I think it is a physical limitation because the Bot is so large.
On one hand, fresh mortar can clearly support several new layers of brick as a wall is laid. On the other, common estimates suggest a 24 hour wait to reach 60% of final strength. How light would an on-wall robot have to be to work at a steady pace without disrupting the still-drying brick?
The other question, I suppose, is whether the brick-supplying robot makes that redundant. It's either going to need a scaffolding or a cart/flatbed that moves frequently to keep feeding target that moves back and forth, and if you need a ground-supported robot arm right near the wall it might just be easier to do everything that way?