The general advice, even from my primary care physician, has been that anything less than 1200kcal/day is dangerous and is crash dieting. Looking around, I found this other study studying total diet replacement providing 810kcal/day as the sole food for 12 weeks, showed significantly greater weight loss compared to standard counseling and modest energy restrictions. The difference in weight loss was measured at 12 months. (BMJ 2018;362:k3760 [1])
25 BMI, which the diabetes study used in addition to T2D, is not considered clinically obese in the US. 25 BMI is the line between clinically normal and overweight. I wonder if crash dieting is advised against on the assumption that most people will go back to unhealthy eating habits instead of changing their diet, and also become more likely to fall into binge-restrict cycles. It's easy to imagine that if someone is able to make meaningful changes, crash dieting is more effective than most other methods.