Damned I was too optimistic :D
So an automatic system is always to find :)

I must make 80 Clickty-clac Zoom of the Mouse Wheel before to see a divergence between 2 arc segments !
Is it compatible with your estimation of 0.1 degre deviation?