Imagine I have a combustion chamber with a nozzle. There is a high pressure oxygen bottle and a high pressure gaseous fuel bottle connected to the chamber. Would the fuel efficiency be lower if I performed combustion in a series of many very short burns than if I emptied the tanks in one go? Since at the end of any burn there's a period where pressure decreases, I imagine that the total impulse of the engine will be lower for the intermittent combustion since it will go through many of these periods instead of just one, leading to increased fuel consumption.
The built-up to momentum in the mass of gas flowing out of the nuzzle and building the design exit Mach and stability of the combustion is better left alone from the disruption of on-off cycles.
Avoiding flame blow out in the combustion chamber and continuity in vector thrust navigation lend better to long bursts.