Hi Finn,
Its been a long time since I dove this deep into cooling ducts/lip contours etc. But, here goes.
The duct (lip + diffuser) is suppose to convert air velocity to air pressure with minimum loss/drag and turbulence. However, like almost everything else involving flight - there is no single best answer - only an optimization or approximation to the best (that you are going to get with the compromises you have to make). So in our case one set of compromises might be between cooling effectivness and cooling drag. As you know the larger your radiator the more heat you are going to get rid of, but, the larger your core the more cooling drag you are going to encounter.
Now at low speeds you might get away with sticking the radiator out in the airstream without any ducting and get adequate cooling, but it will have higher drag - now at biplane/slow airspeed the drag might be acceptable, but at 200 mph quite unacceptable. Again, it depends on what you are trying to achieve.
Its been quite a while, but Ill give it a shot on trying to explain the charts. I presume that External refers to air factors prior to the lip and Internal refers to air factors after the lip of the diffuser (or internal to the duct). you mention wanting a 100-20 mph slow down. That is a data point, but as I recall that is the overall slow down which would include the slow down across the lip of the duct and the across the radiator combined. Also for a faster aircraft where Vo is quite high, perhaps Vi/Vo = 0.40 makes for less drag but gives you all the cooling needed as opposed to selecting cooling as a Vi/Vo = 0.20.
Also the Positive CPs are all below the horizontal line for Internal flow. With lower Vi/Vos resulting in higher positive CPs. That makes sense because if we have Vi/Vo = 1.0 then that would imply the least conversion of velocity to internal pressure, and we get a negative ratio Cp = -0.6. Conversly if Vi/Vo = 0 then that would imply all the velocity energy is coverted to pressure Vi/Vo = 0 then Cp = max or approx positive 1.0 which tends to jive with the graphs.
The Cp external magnitude tends to increase as A (lip Curvature) becomes more pronounced (with A=40 being the most pronounced and A=10 the less pronounced lip curvature). External pressure would be considered a drag component in my opinon, so you want the least of that possible consisten with other objectives (such as cooling).
So I think A=20 was chosen as the best compromise of lesser drag and adequate cooling. Also, other lip curvatures might have had "Separation" or other problems at different angles of attach (from 12 to -12 degrees). For example perhaps if A=40 was choose, separation may have occured at angles below 12 Degs and for that reason alone been unacceptable.
Best this old brain can do perhaps someone of the younger generation can jump in and clarify.
Ed