Hmmm... The very high feedback of many (most?) class D amps provides lower distortion and lower output impedance than the class AB amps I have glanced it over most of their power and the audio frequency range. Not saying class AB is audibly worse, but the measurements seem to show an advantage for class D.Exactly! If you manage to put a class A/B into a well done composite configuration, it outperforms class-D in almost all aspects up to say 250 Watts output power!
I have not seen any hard data on power dissipation comparisons at low power levels; may have missed them (busy day).
I do find class D's high-frequency distortion, even when bandlimited to well below the self oscillation (switching) frequency, curious. I would have expected loop gain to be high enough to stay competitive or better than class A/AB amps well beyond 20 kHz, unless HF artifacts are being aliased/mixed back into baseband. @pma's data are very interesting and the first I have seen in a while for audio amplifiers. I have not tried to research it though believe Bruno's papers may offer insight (?) Times like these I wish I had kept my AES membership.