Set Roman pupil radius correctly, even if pupil files have zero padding

[mperrin]

This should fix #151 reported by @bergkoet.

Currently, we are setting both a RomanTelescope.PUPIL_RADIUS class attribute (~correctly) and a self.pupil_radius instance attribute (incorrectly), which is both not working and also more complex than is necessary. This PR simplifies to just setting a self.pupil_radius value, correctly. Conveniently the PUPIL_RADIUS one was only used in one single place in the code so this is not a complex edit.

I moved the numeric value definition to constants.py which is a better place for it anyhow.

[bergkoet]

Thanks for the fast response!

This doesn't quite fix it because the .calc_psf method has the side effect of changing the pupil radius (by way of .get_optical_system). It happens here:
https://github.com/mperrin/stpsf/blob/dfad7477ffbb330177c7be792d1234063df4f6b4/stpsf/stpsf_core.py#L450

Here's a minimal example:

import stpsf

WL = 1.577e-6         # F158 central wavelength [m]
DEFOCUS_WV = 0.25     # requested defocus [waves P-V]
EXPECTED_RADIUS_M = 2.36312 / 2  # physical entrance pupil radius [m]

stpsf.setup_logging()
wfi = stpsf.roman.WFI()
wfi.filter = 'F158'
wfi.options['defocus_waves'] = DEFOCUS_WV
wfi.options['defocus_wavelength'] = WL
found_radius_before_m = wfi.pupil_radius

psf, wavefronts = wfi.calc_psf(monochromatic=WL, return_intermediates=True)

found_radius_after_m = wfi.pupil_radius.to_value('m')
rho_max = EXPECTED_RADIUS_M/found_radius_after_m

print(f"\n")
print(f"Physical entrance pupil radius:   {EXPECTED_RADIUS_M:.4f} m")
print(f"wfi.pupil_radius before calc_psf: {found_radius_before_m:.4f} m")
print(f"wfi.pupil_radius after calc_psf:  {found_radius_after_m:.4f} m")
print(f"rho_max = physical / ThinLens radius:    {rho_max:.4f}")
print(f"Expected defocus underscale (rho_max^2): {rho_max**2:.4f}")

Result:

Physical entrance pupil radius:   1.1816 m
wfi.pupil_radius before calc_psf: 1.1816 m
wfi.pupil_radius after calc_psf:  1.2763 m
rho_max = physical / ThinLens radius:    0.9258
Expected defocus underscale (rho_max^2): 0.8571

[mperrin]

@bergkoet OK let's try this again. Thanks for the minimal example code, and for pointing out the exact line causing the error; that was very helpful. With the latest commit just added to this branch, I have your test code passing:

rho_max = physical / ThinLens radius:    1.0000
Expected defocus underscale (rho_max^2): 1.0000

[bergkoet]

@bergkoet OK let's try this again. Thanks for the minimal example code, and for pointing out the exact line causing the error; that was very helpful. With the latest commit just added to this branch, I have your test code passing:

rho_max = physical / ThinLens radius:    1.0000
Expected defocus underscale (rho_max^2): 1.0000

That works for me, too. Many thanks!