the 810 and 1050 specifically, ran at 288 RPM rather than 300. I can't imagine why this might be,
I do believe it was needed to securely store 18 sectors per track while using the WDC1771 at 1 MHz feed by a 6507 at 500 kHz. It might have been a sweet spot for components, with slowing down the drive (zero part cost) enabling some savings.
Worth noting that the XF551, Atari's last 5.25, did indeed run at 300, which I believe was because they used the drive's built-in controller.
No. It was done by simply upping the master clock from 8 MHz to 8.333 MHz, which is almost exactly the 4.1% speed increase from 288 to 300 rpm.
This did cause minor problems with some disks, but nothing serious.
Not really. With the changed master clock differences were way within the original specs. Genuine Atari floppies suffer only with early 810 drives which were not really stable in their analogue parts.
Then again, rotational speed is an issue when disks formatted on a 810/1050 are written on non XF551 300 rpm drives (like on a CP/M machine, or PC), as they may write over the header of a following sector.
So it seems one could have used 300 at any point.
If they had increased the clock speed of the CPU/controller system.