Surely a PWM of 10,000Hz wouldn't allow for night sky photography. With full exposure it can already take minutes of hours to take a night sky a photo.
And it's usually recognized that 1,000hz is the bare minimum to spare sensitive individuals from symptoms, with 2,500 usually used for safety margin. Then you have to account for some jitter in the synchronized network so you're looking at at best a 1/5,000 exposure.
Is that sufficient to be worth the trouble?
You can still get a long exposure by accumulating many 1/5,000 exposure shots together. Just like a regular long exposure, this will still cancel out random noise and improve SNR, and also still has the tradeoff of motion blur. This kind of accumulation is commonly done in photography-adjacent technologies like spectroscopy.
Regardless, I agree with your larger point that the proposed system would be entirely impractical!