A couple of thoughts:
Missing from the discussion so far is the limitations imposed by the (CCD) detector.
To extract all the information at the focal plane you will need to oversample the image by at least a factor of 2x
(basic nyquist sampling theorem)
However, the following caution also needs to be included:
Do not assume that a CCDs sampling rate is defined by its pixel size.
A real world example of what I am alluding to can be found at the bottom of page 11 on the following PDF:
http://www.sta-inc.net/wp-content/up...-Astronomy.pdf
In this example, when a 2 micron star image was focussed on to an 9 micron array, 90% of the image was
registered by the target pixel, the surrounding pixels acquired the bulk of the remainder, rendering
it as a 27 x 27 micron cross. (13.5x the size of the original image) This is not due to the central pixel's well
depth being exceeded.
Implicitly therefore, for mirror quality to become a significant limitation in deep sky imaging, you need to
be over-sampling to a significant extent. The magnifications used by planetary imagers might serve as a data point here.
The bottom line is that if you put an ST10 at the prime focus of an 8" F5 GSO Newtonian equipped with a coma corrector,
the resulting spatial resolution in the image will be indistinguishable from what you would get with a 5" f8 Astrophysics
refractor, all else being equal (not withstanding the diffraction spikes & better image depth with the GSO).
best
~c