The variation in voltage between the legs, caused by unbalanced 120V loads and resistance in the neutral conductor, is something that keeps coming around. I took a hard look at this quite awhile ago, so it’s time to revisit that:
I did take my own advice and have my production IoTaWatt setup with the 230V Euro transformer connected to my 240V. I scale down the calibration to 120V (actually calibrated to half the 240 voltage). When doubled for the 240V loads, it’s always exactly right. More than half of my usage is from 240V appliances. The total of the error for the 120V legs is also reduced significantly.