Prog, you're wrong. You put a voltmeter in parallel with what you want to measure the voltage of. When modeling an ideal voltmeter you assume infinite resistance. Real voltmeters have some real resistance, but the current flow is minimal, so he's not really shorting the batteries with the meter. Unless you are saying a ~.1 microAmp current to constitute "shorting." Now, if his mutlimeter probes were shorting with each other that would, of course, short the batteries. But I didn't see that happen. Plus things would have gotten hot and I'm sure he woulda noticed something was wrong.
If he was measuring current then yeah, that's in series, but he was measuring voltage. Hence the "V" seen in the video.
As for why it's doing this, I can't really be certain without a better look at everything.