There’s a lot more factors at play here - like local upwash/downwash, aerodynamic and geometric twist, wing sweep etc.
I’d argue that even on a plain, straight wing with constant angle of incidence, each section operates at a different local AoA, because of 3D airflow. Therefore it’s sensible to pick up a different reference.
Also, in any case, the reference does not matter, as long as it’s consistent and data is appropriately adjusted for it. For example, I seem to remember from reading in one incident report (probably the Qantas A330 upset), that A330’s AoA vanes ‘zero’ reading was something like -45degrees down from the fuselage axis