I am not for one moment disputing how much more dangerous performing a dry/air test compared with one using water, however to claim that there will be more pipe/joint expansion when using air at the same pressure as when using water is ridiculous.
Just what I was thinking.
The biggest advantage I can think of for wet testing is that the weight of the water will make the pipe less buoyant, and so it will be easier to keep it properly down while the screed is poured. But once the screed has set then that's not a consideration. Common sense suggests that if there's a risk of frost then you either need to empty the pipes (blow them out with air) or put antifreeze in. Blowing them out isn't guaranteed not to leave a low spot that the water will run to and still leave you with a section of pipe that's completely filled and liable to freeze damage.
It can't be an uncommon problem in winter - at least in some areas.