It's a good question. The thing with autorotation is, it does not work very well when considered from a standpoint of vertical flight - it's not really a windmill. The best example for explaining the principle of autorotation is the autogyro: it always flies in autorotation, but never vertically. Or at least, never very successfully for long. From Principles of Helicopter Aerodynamics by J. Gordon Leishman:
The picture illustrates a pretty important issue: it identifies lift and drag forces in forward flight. If the autogyro or helicopter would descend vertically, the rotor would create drag only, since drag is defined as the force in the direction of airspeed. Increased drag, like a parachute. The rotor would act like a partially permeable fixed disk. That would not work very well for stopping the autogyro or helicopter from falling.
It is much more effective if the rotor behaves like a wing: forward flight creates a lift force, and drag is overcome by a propeller or by energy absorption from reduction of potential energy. In other words, if it behaves like a glider. Gliders don't glide very successfully in vertical direction either...or at high angles of attack, which is what is indicated in your drawing at 45º. The rotor blades are wings, but sadly with difficult dynamics - please hang on.
In autorotation we have rotating blades, one going forward and one backwards. The forward going blade experiences headwind, the backwards going blade has a tailwind. The rotor is driven by difference in drag, and if the forward going blade experiences less drag than the rearward going blade, the rotor is driven by the wind force - it never changes direction of rotation. That sounds counterintuitive, but it is exactly what happens.
When the first autogyro was put to a flight test (way before the first helicopter, in early 1920s) it started rolling when forward speed picked up, because the forward going blade has higher relative wind speed and therefore more lift. The first successful flight, of the C4 in 1923, fixed this by introduction of blade flapping, really quite a brilliant design. From Wikipedia:
Dissymmetry is countered by "blade flapping": rotor blades are
designed to flap – lift and twist in such a way that the advancing
blade flaps up and develops a smaller angle of attack, thus producing
less lift than a rigid blade would. Conversely, the retreating blade
flaps down, develops a higher angle of attack, and generates more
The same blade flapping creates a driving torque of the rotor, by difference in drag between forward and rearward facing blades. But it only does that when the collective is fully down. The collective flight handle sets the Angle of Attack of all blades simultaneously, and the first thing the pilot needs to do when the engine stops is to lower the collective. Otherwise the drag in the forward going blade is higher than in the rearwards going blade, and the rotor stops turning.
So autorotation only effectively works with forward velocity. In the last bit, just before landing, the pilot does go almost vertical: he raises the collective, trading rotational energy of the blades for loss of kinetic energy, and he comes to a beautiful soft landing.