The relevant angle is the projected angle, which Tomasz defines in section3 and Figure 2. 
The projected angle is always smaller than or equal to the angle through which the sensor 
is twisted, and strongly varies across the sensor.

Figure 12 shows the resolution as a function of the projected angle.This plot is for 
simulation, the previous testbeam data matches well with these values.

We would be primarily interested in measuring the resolution for the region with the 
smallest pitch (i.e. best resolution).For this region a 20 degree maximum projected angle 
would be fine. As you see a separation of 5 degree points in projected angle would be OK, 
but somewhat smaller would be ideal to follow the shape.