PBL,
yes, and they do indeed go on using a ground speed as given by a doppler radar to derive Vnorth and Veast. Ground speed given by a doppler radar would in fact (more or less... subject to inherent errors and inaccuracies in the system) be the rate of change of position of the aircraft as projected orthogonally onto the geoid, as per my preferred definition above.
This also agrees with the initial use of ground speed for a ground-bound vehicle.
For all flights which are not
very local in nature, I suspect that we can happily ignore the geoid vs. ellipsoid debacle as the errors introduced on the momentary ground speed will even out.
Interesting, this!
