As someone in the business of Air Traffic Management research for over 20 years, I have to say that the constant descent, engine idle approach is very much an ATC system issue rather than a piloting issue. It is in fact the 'holy grail' of ATM research.
The flight crew could probably get it spot-on most times - maybe every time with the help of a computer to accurately nail the TOD.
The secret is to only let one aeroplane into the sky at a time. More than that and they have a nasty habit of interacting in a 'safe separation' kind of way. When there are hundreds of aircraft up there the interactions become chaotically complex - not in the sense of being dangerous, but in the sense of defying any truly analytical solution to gaining the maximum efficiency for each and every flight.
Many attempts have been made. Much faith was put on 4D trajectories through the 90s. However any speed adjustments made more than an hour out have a habit of being rendered pointless by some unforseen interaction in the last 20 minutes.
So far, the only really succesful solutions for well spaced arrivals tend to be based on path extension (and even here the computer assistance tools may provide only a marginal improvement on old fashioned ATCo skill). And this is of course totally contrary to the aspiration of constant / idle descent.