Sorry about the delay in my own reply.
I haven't checked to confirm, but I would imagine that a single Arduino could do the required calculations for one thousand heliostats in hardly any time at all. Maybe in just a few seconds or less?
The real delay would be in sending the info wirelessly, although even that should be pretty fast, assuming you are using XBees.
You may have already seen this, but here is someone who has done something that sounds similar to what you want to do.https://www.youtube.com/watch?v=pwgBR2O9nr4
Here is a pdf on the project. http://digitalcommons.calpoly.edu/cgi/viewcontent.cgi?article=1507&context=theses&sei-redir=1&referer=http%3A%2F%2F
I know of two other projects that use cameras for tracking. One was by Google and the other was a commercial heliostat who's name I can't remember right now. In both cases the tracking doesn't actually seem to be especially good.
I don't understand how you would manage multiple mirrors with a camera system. I mean, if the reflection for one is off, how would the software know which heliostat needs to be moved? All the reflections look the same.
(OK I just reread your post. You're only trying to double check the position of the sun aren't you. The above doesn't really apply.)
The laser and photoresistor idea might not work for what you need, but basically one Arduino would turn the laser on and off. The light from the laser would shine on the photoresistor which is connected to another Arduino. Assuming that the flashes sent a binary code, the second Arduino could use them to determine how many steps to move each motor. Of course, you wouldn't be able to have anything in between the two Arduinos though since it would block the light.
Another option might be this. http://www.glacialwanderer.com/hobbyrobotics/?p=291
I haven't experimented with it myself, but it might be your best option. Probably better than the laser pointer.
The NEMA 17 motors should be fine, depending on their torque. The torque can actually vary greatly from one type of NEMA 17 motors to the next. You're basically looking for the high torque NEMA 17 motors, not the high speed motors.
The azimuth and elevation axis does not need to pass through the center of the mirror. It's been awhile since I did it, but I checked for errors using a CAD program and by testing in real life, and there didn't seem to be any errors added by moving the mirror away from the center.
Hope that helps!