I don't know who has written that in dynamicflight.com but it's all wrong, big time, both the picture and the quoted text. Obviously written by someone who has no clue about helicopter rotorblade azimuthal flapping.
Lets assume the flight controls (at least cyclic) are in neutral. In hover all blades have the same angle of attack and flapping, that's clear. In case of a forward flight (or any other direction of the velocity vector) and counter clockwise rotor rotation (like in that picture) the advancing blade at 3 o'clock will have the biggest average velocity hence biggest list (depending on V^2) and biggest flap up which decreases the blade AOA. It's called selfcompensation or blade flapping self-damping. Because of the blades inertia (like any other body) the maximum angle of flap is indeed not at 3 o'clock (biggest lift) but phased forward up to 30-45 degrees. That 90' (azimuth 0-180) on that picture is a joke. Similar is what happens on the left side with the retreating blades. At 9 o'clock the blade will have the smallest airflow velocity which will decrease the lift and force the blade to flap down thus increasing it's AOA, hence the retreating blade stall nature. Again, thanks to blades' inertia the lowest point the blade reaches is not at 9 but about 8 o'clock.
As a result, in that case the rotor cone in forward flight will be tilted to the left and a bit backward. And that's why ALL helicopters around the world have their rotor (gearbox) axis tilted forward- when in cruise airspeed the rotor lift should not have rear component.
As for coaxial counter rotating rotor the picture is all the same for any of the two rotors separately. Two lower rotates counterclockwise (seen from above), the upper clockwise. In forward flight the lower rotor cone tilts to left and back, the upper to the right and back. This makes the gap between the two rotors to decrease at the right side, between 2 and 4 o'clock (helicopter nose is 12).