Correct me if I'm wrong :
When using lateral and vertical type mosfets in output stages the Vbe multiplier is left off of the heatsink. This wasy it doesn't track the heat and just remains an amplified diode. Thus as the OPS heats up the transistor in the Vbe multiplier retains ambient temperature (I.E. not OPS temp) and consistant forward bias. In such a case, X-over distortion should not occur as the OPS heats up (and stabilizes). If using D-MOS such as the IRF devices, I think that the Vbe multiplier needs to be in contact with the heat sink as would BJT's to prevent thermal runaway.
I've seen many schematics that use both M and V - MOS with a Vbe multiplier, but generally there is little documentation of heatsinking. I guess most designers are well versed in the subject and assume the same of anyone reading the schematic.
It should also be noted that different MOSFETs will require different voltage biases. The Lateral types needing the least amount with the IRF types needing the most. The V types seem to be in the middle, tho closer to the L types.
I like the idea of using a Vbe multipier as this gives MUCH more contral over the biasing than using any sort of diode. I also think that it sounds better because of this control.