As an electrician, if I were auditing such a system, I would not trust a wall receptacle. Assuming I knew the expected load, I would run cabling from a panel as close to the building's source as practicable, and supply my own breakers. I've put a lot of wires in walls and had them covered. Who knows what's in there after the drywall is on?
Meters like the Kill-a-watt plug directly into the wall socket, with a standard 3-prong plug. The device you are testing then plugs into the Kill-a-watt. So I do not see how hidden wires could "get around" it. All of the electricity has to go through it, and as I said, it will either measure the electricity or burn up. If there were hidden wires the only thing they could do is supply more amperage than a standard wall socket, and the only thing that could do is burn up the plastic Kill-a-watt gadget. It isn't big or rugged.
You could also use a computer battery backup for this purpose. Some of them have built-in volt-meter or watt-meter functions. With a device plugged into the "battery backup" sockets, I do not think it is possible for one of these to supply more than the rated power shown on the back. The "surge only" sockets might produce more than the rated power.
You can also unplug a battery backup and watch the battery drain. I have an APC battery backup that has an optional connection to a computer, so you can collect data from it. At this moment it shows: 121 V, 122 W, 60 Hz, 39 minutes of battery power remaining.
A Kill-a-watt or similar watt-meter might not produce definitive proof that the Rossi device is (or is not) producing more energy out than in, but it would produce far better evidence than his test did. It should be combined with other instruments and test methods. The Kill-a-watt should have been supplied by one of the audience members, and plugged in the first time just before the test. I think there is no way Rossi or anyone else can alter one of these gadgets, but there is no point to leaving that possibility open.