That paper says that "in no case were equal numbers of photons used more efficiently". Now, when you pulse, there's going to be less photons in total than continuous.

The conclusion tries to tell you that pusling is not as good as continuous, and the idea that it is stems from the misinterpretation of the LUE. Yet, devilgoob saw 2700% and misinterpreted it.

The final sentence summing up the whole thing starts with "although there is no advantage to providing photons in pulses".