They're spaced out over miles, so presumably they'd take a wide shot of a bunch of them and check that the focus on one side of the frame to the other would be consistent.
They took a whole series of images at different focuses, then dropped it to earth and people radioed back the most in-focus setting....is how I'd do it.
Well, I'm going to assume there was no actual "computer" on board...
However, a sharply focused image has more high-frequency components than a poorly focused one, so some sort of frequency discriminator (a technology from the 1920's IIRC) could possibly suffice to check and adjust focus. Remember that we've had sophisticated servomechanisms since WW2 so the feedback theory to do this was already well known.
Think of it as an eye chart for satellites.