Of your amplifiers, I'm only familiar with the circuit in the GAS amps . . . and both are fairly representative of a common-practice solid-state amp with an emitter-follower output. The complimentary diff-amp is kinda fancy but by no means unique, and the Ampzilla's bootstraped outputs are typical of many higher-powered amps of the era.
So to answer your question, small amounts of Vos are really only possibly of concern to your woofers, in terms of a static displacement in the voice-coil position or eventual loss of magnetic flux. While there is indeed a small difference in current between the halves of the output stage as a result of a DC offset, the optimum bias point in a Class B amp is dependent on the voltage between the output transistors' emitters, not the current through the output stage. And yes, they are sometimes the same thing in practice, but if you were to change the value of the emitter resistors, the optimum bias current value would also change. And since the bias generator sets the voltage between the two halves of the output stage (not relative to ground), the presence of an offset doesn't change the transfer function of the output stage.
The main issues with output stage biasing are how well it's maintained with variations in temperature and line voltage, and how much an increase in signal current affects the linearity in the crossover region. And these issues span everthing from circuit design, the characteristics of the transistors themselves, the layout of the wiring and circuit board, and the physical packaging and thermal design . . . there are no universal solutions.
