The mirror is a pretty good sphere, a bit rough from the machine polishing but nothing that is causing the problems in the star test. So, that leaves the secondary and the corrector for causing the overcorrection. A optical flat has no optical power so it can't add spherical aberration. If the surface isn't flat it will add astigmastism. So that leaves the corrector like you said that is the problem.
Tim's sugguestion to do double pass autocollimation is good one. That tests the complete system. There is another test you can do to check the optical smoothness of the corrector. You can place it a few inches in front of the spherical primary and Foucault test the assembly. It doesn't matter which side faces the mirror so you can flip it around so the diagonal is facing away from the mirror If the corrector was perfect you would see smooth surface that tests like a parabolic mirror ie showing over correction but again you'll get an idea of how smooth the corrector is. It is best to test it with Ronchi screen since the bowing of the lines is easier to understand what the shape is and if there are problems.
A note on your Foucault test set up. It is better to have the light source and return image as close as possible. If not your introducing astigmatism into the results because you are off the optical axis. So you should raise up your frame so the LED is close to the center of the camera lens. Also I can't tell from the picture but the knife edge should cover about 1/2 the LED. Again this puts the light source and the return image closer to each other and closer to the optical axis.
Here is a picture of corrector I tested this way. It shows astigmatism from "S" shape of the ronchi band in the middle