This is directed principally at Frank (freestar8n)...
The singular reason that a sufficiently point-like image of an axial light source produced at the *common focus* of eyepiece and objective requires no knowledge nor consideration of the system pupils is this...
The envelope of light passing back-to-front through the system is simply the reverse of that light accommodated by the system from a distant point source placed at the field center. Collimated, axial light enters and emerges.t Whether the offending obstructor be the corrector perimeter, the primary perimeter, the secondary baffle, the primary baffle, or some other nearer-to-focus stop, the emerging light bundle diameter in the flashlight (or, more reliably, the BELT) test will match the diameter of the entrant cylinder of light from a distant point source and which makes it through unobstructed out the eyepiece.
The consideration of pupils comes into it *only* when dealing with or including the *off-axis* condition. The flashlight/BELT test confintes to only that singular, axial light cone between the collimated entrant and exit bundles, the latter of which rays are all paraxial.
The simplest sketch shows this. Diagram a basic refractor with eyepiece, with paraxial rays (for the full aperture) passing through the system--parallel entering and parallel exiting. Then place at any arbitrary place between objective and focus an aperture stop of any diameter. Draw in the new envelope of rays that can pass through the system. We see *allways* the case of precisely similar triangles. Whether the obstructor be at the objective, a little behind the objective, far behind the objective, or pretty near to the focus, if at its location its aperture relative to the light cone diameter there be, say, O.7, the instrument in all cases works at a relative aperture of 0.7.
The same applies for far more complicated systems that have entrance pupils in 'strange' locations. Our setup is concerned with only the same paraxial rays entering and exiting, with the single, axial light cone between. No matter where the aperture-limiting obstructor is located, its impact is no different than just described for our simple refractor.
The beauty of the flashlight/BELT test is that we are using the system itself to do the 'ray tracing'. As long as our light source appears to the system as subtending a suitably small angle, the emerging bundle will be collimated to sufficient degree for some confidence in the result. For example:
Suppose or instrument under test has installed in it an eyepiece delivering 100X. Our light source has a diameter of 5mm, and is placed 200mm from the eyepiece. What is the divergence of the emerging light (i.e., its departure from parallelism)?
As 'seen' from the eyepiece, the light subtends
ATN(5/200) = 1.43 degrees
The eyepiece 'de-magnifies' the image as 'seen' from the objective by an amount equal to the magnification. And so the light emerging out the objective has a divergence of
1.43/100 = .0143 degrees.
In the normal direction of light travel during observation, then, an object of diameter 0.0143 degrees would be seen with this 100X eyepiece to appear to subtend 1.43 degrees.
We see that with a modicum of care in light source size (or characteristics) and its placement, we can synthesize what is effectively a quite small point of light at the *common focus* of objective and eyepiece. The BELT test, by use of a laser of already darned fine collimation, yields a truly diffraction-limited source that suffers only from that phenomenon of diffraction.
Finally, in this specific setup where only the *axial* light bundle is of concern, please disabuse yourself of the concern about the quite unnecessary pupils. Where the light under consideration is axial and collimated at both entrance and exit surfaces of the system, the pupils can be said to lie *anywhere and everywhere* along the optical path. Because the field size is essentially zero. Because there are no crossing off-axis rays by which to locate a pupil *anywhere*. The pupils have effectively expanded to the length of the envelopes of paraxial light bundles, as strange as that might seem.