Short answer to “is there any advantage to” is depends. Short answer is “and technique to allow” is yes.
This is a never ending question with no definitive resolution. I believe it all started on another site where there are academics who beat it into the ground with theory, but introduce no practical evidence.
Here’s a post that explains my thinking on this topic: 120V/240V split phase voltage reference
Where you have two VTs in close proximity, if you wanted go to the trouble of wiring them to opposite legs and splitting them with CCTV type voltage splitters that observe polarity, and sacrifice a power input on each device to bring in the additional voltage reference, you could assign each 120V circuit to the corresponding voltage reference and be more exacting with those. But the bulk of the power in a US service is typically consumed by 240V appliances like water-heaters, AC, and clothes-dryers. This does nothing to address that. When the voltage of either leg varies from median, the other does the same in the other direction. So the doubled 120V reference for a 240V circuit is off by +/- twice the voltage difference. That’s why my final recommendation in the above post is to use a 240V VT and calibrate to the median. It’s simpler and more effective than using two 120V VTs.
All that said, I have a 240V VT yet I haven’t bothered to switch from the 120V VT which has been giving me very accurate results for years - within 1% of my revenue meter.