Well they might just give me back my "King Measurebator" title for this (not that I want it).
After dealing with a sync speed issue with my Elinchrom Skyports, I began to wonder how much an optical slave might delay the flash it's attached to and make such an issue worse. So I decided to test it.
This shot shows my ceiling being illuminated by two flash units - The one on the left is triggered by the Skyport radio set and the one on the right is attached to an optical slave. So in this setup, the flash on the left is the "master".
When the shutter speed is increased beyond the camera's X-sync speed, the flash fires while the second shutter curtain is closing. The dark area in the bottom of the frame is the part of the sensor shaded by the shutter curtain.
Notice that the height of the black area is nearly the same on both sides of the frame. This indicates that the flash connected to the optical slave fired almost simultaneously, even though it didn't fire until it "saw" the other flash. If the second flash fired with any significant delay, the black bar would be higher on that side of the image.
The difference between the last two images represents the distance the shutter curtain travelled in 0.00625 seconds (1/1600 second). The nearly imperceptible difference in the height of the black area on either side of the frame indicates that the delay of the optical slave was less than 1/10,000 second.
What does all this mean? Well, it means that an optical slave will not cause a reduction in the practical maximum shutter speed with flash, even if you're using a camera with an electronic shutter that can sync at 1/1000.