the drag and moments are much more sensitive to the angle of attack

Too true, Jean-Claude, the whole procedure stands or falls with the accuracy of your calculated rotor wake. Fortunately today's Computational Fluid Dynamics (CFD) programs are extremely powerful. Time and again I am amazed at what kind of problems our CFD chaps are able to investigate successfully (I'm currently working at a Siemens steam turbine plant). It is therefore today no problem to calculate the angle of attack of any fuselage part immersed in the rotor wake to great accuracy. (of course my program is hopelessly outdated in that respect, using analytical expressions for the rotor wake)

(why) lift, drag and moment coefficients are linked to the radius of a ghost rotor

As you see in the picture of the 139 the fuselage is cluttered with all sorts of things: a rescue winch, an antenna running the length of the tail boom, something looking like a mount for a depth charge and so on and so on...

Therefore in helicopter design the frontal area of the aircraft is almost never used, instead the drag is calculated based on the so called "flat plate area". This basically means that the frontal area and the drag coefficient of the aircraft are combined into one number. For a fixed wing aircraft the drag force is:

fD= 0.5*rho*cDfus*Afus*V^2

where rho is the density of the air, cDfus is the fuselage drag coefficient and Afus is the fuselage frontal area. V is of course flight speed.

For a helicopter that almost always is

fD = 0.5*rho*fPA*V^2

here fPA is the flat plate area and is equal to cDfus*Afus

As we have seen the aircraft drag depends on various parameters that might change so it is a good idea to normalize coefficients with respect to rotor diameter, since that one does not change and is readily available.

I am currently struggling with the Lynx coefficients and would be very happy if these were normalized with respect to the rotor diameter, unfortunately they are not...;-(