The resolution of an optical telescope can be calculated using the Rayleigh criterion: this is the theoretical limit: an optical system cannot be better than this.
$θ=1.22λ/D$
θ is the angular resolution (the smallest angle between two objects at which they can be resolved as separate objects), λ is the wavelength of the observed light, D is the aperture of the telescope (basically, the diameter of the mirror).
if we pick λ is 500 nm (green, ~ in the middle of the visible spectrum), D is 100 m, we get $6.1E^{-9}$ radians.
Now we convert that angular size to a linear dimension on Earth's surface:
$A=2L*tan\frac{θ}{2}$
L is the distance, 36,000 km. This gets us a linear size of 0.21 meters. That's pretty close to the resolution of low-altitude spy satellites (KH-11: theoretical resolution of 0.06 m, in practice it's worse due to atmospheric distortion).
Then we get to the practical problems:
The required accuracy of the mirror depends on the wavelengths you want to observe at. So the radio antenna on the Orion sat can have variations in shape on the order of centimeters, while an optical mirror has to be accurate to within tens of nm. For radio, you can get away with an umbrella-type unfolding mechanism with a mesh in between the ribs. An optical mirror has to be thick, accurately ground, and placed on a frame with actuators so you can adjust its curvature. The largest mirror in space is on JWST, at 6 m diameter.
You could segment the mirror to make it easier to launch, but that adds lots of optical noise (diffraction spikes, ie light that reflects off the edges of each segment, this shows up as spikes if you use e.g. hexagonal mirror segments, or an Airy disk if you use circular segments). This reduces the resolution.
To create useful images, most of the surface area has to be reflective. A radio antenna can use a mesh, if the holes in the mesh are smaller than the wavelength of the radio signal. For optical mirrors, it's not practical to make a mesh mirror. You can use a segmented mirror, see point 2.
