FFT diffraction doesn't include circle sources from the geometry

[welucas2]

Issue #506 covered a bug in imsim/diffraction.py which caused suppression of diffraction of photons by circular structures, the circles in RUBIN_SPIDER_GEOMETRY:

imSim/imsim/diffraction.py

Lines 32 to 42 in f20cecf

	RUBIN_SPIDER_GEOMETRY = Geometry(
	thick_lines=np.array(
	[
	[1 / np.sqrt(2.0), 1 / np.sqrt(2.0), -0.4, 0.025],
	[-1 / np.sqrt(2.0), 1 / np.sqrt(2.0), -0.4, 0.025],
	[1 / np.sqrt(2.0), 1 / np.sqrt(2.0), 0.4, 0.025],
	[-1 / np.sqrt(2.0), 1 / np.sqrt(2.0), 0.4, 0.025],
	]
	),
	circles=np.array([[0.0, 0.0, 2.558], [0.0, 0.0, 4.18]]),
	)

This was due to the incorrect normalization on vectors between photons and the circles, leaving visible only the spikes caused by diffraction by the lines. Fixing that (#507) adds a disk of photons, which didn't occur previously.

Bright objects which are usually handled with FFTs model diffraction through different code in imsim/diffraction_fft.py. This only includes the spikes, i.e. the diffraction from the thick_lines component of RUBIN_SPIDER_GEOMETRY, and neglects the new disk we saw from the circles. Diffraction from circles in the geometry should be added to FFT diffraction.