It's my understanding that the ISO on a digital sensor is determined by how much electricity the sensor is charged with. Where is the information coming from that says the intermediate stops are 'pushed' or 'pulled' with software? It wouldn't make a whole lot of sense to design a processor to manipulate an image when the same processor can simply increase a voltage parameter to the circuitry feeding the sensor, thus increasing or decreasing the ISO.
It may be software that calculates the voltage required for each ISO selected, but overall there should be no difference in how the sensor receives it's voltage. Basically what I'm reading here is that for Native ISO's, the camera has circuitry to send a certain voltage directly to the sensor, while the intermediate ISO's process the image differently. I can't get my head around the possibility that Canon doesn't simply line up a few more tiny resistors, or capacitors to achieve the ISO, but instead uses software and processing power at the processor to achieve the same thing.