They used idle power to measure the difference between going idle vs. full tilt. What else would be drawing more power other than the SOC during power intensive tests? Is the Mac Mini's power supply that inefficient?
Intel's docs also say that on chip power monitoring (which is what Max only relies on) is not as accurate as just measuring power draw differences at the wall.
Intel's claim makes no sense, because power draw from the wall would also include the power drawn by the screen (if using a notebook), the HDD/SSDs inside the machine, RAM, WiFi/Bluetooth cards, etc.). If the processor is running a heavy load, data swaps to and from both RAM and storage would increase as well, which would increase the power consumption of those components.