As for sizing the resistors, you need to consider the DC bias drawn by the stage and the resulting voltage drop through the resistor. The power supply voltage must be made larger to accommodate for this voltage drop, and/or the resistor needs to be made small enough to result in sufficient supply voltage after the isolator.
An opposing requirement is that it's generally good to make the 3dB frequency of the RC filter below the audio band, probably several Hz if you can use large enough capacitors. The numbers you show above are good general ranges - 470µF with 100-150Ω will be outside of the audio band, and thus will filter transient currents out of the supply rails (and move them to the ground point of the shunt capacitor).
As for why you'd do this, there are a number of reasons. One is to move transient currents out of the power supply and into the local circuit ground, which will tend to make the signal current loops in the circuit smaller, and thus less troublesome to other circuits. It also allows you to essentially ignore the power supply impedance, and to some extent, power supply noise as well, since the RC isolators will filter out some power supply noise, and the shunt capacitor impedance alone will determine the effective supply impedance over the audio band.
Another reason why this could be useful is if each amplifier or amplifier stage has very low (or no) power supply rejection. This is the case with a common cathode tube amplifier stage - plate voltage variations can go straight into the output of each stage with little attenuation, making some filtering and isolation worthwhile. So, filtering schemes that start at the output stage, and provide a chain of RC filters that work their way to the input stage, will make good use of the voltage drops and provide the greatest filtering and isolation at the first stage where there is little feedback to counteract power supply variations.
With op amp circuits, which have extremely high power supply rejection, you may not want to do this at all, and instead rely on a regulator to provide a low impedance supply, which will inherently have isolation between stages because of its low impedance. The transient currents will then be kept out of the ground system, and will stay in the power supply rails, and this can be good or bad depending on the overall design.
I hope this helps!