Support beta approximation for models with local structure factor

[pkienzle]

A number of models are localized and so can exist in various concentrations, but are not defined in terms of a simple squared form factor F. Without F these models cannot be used directly in the beta approximation calculations. Need to determine how to support non-dilute solutions of these models.

They are:

flexible_cylinder
flexible_cylinder_elliptical
linear_pearls
mono_gauss_coil
pearl_necklace
poly_gauss_coil
polymer_micelle
pringle
raspberry
stacked_disks
star_polymer

See #180 for prior discussion.

Maybe support lamellar models as well.
#173 (comment)

Models with long range structure factor:

binary_hard_sphere
bcc_paracrystal
fcc_paracrystal
fractal
fractal_core_shell
lamellar
lamellar_hg
lamellar_hg_stack_caille
lamellar_stack_caille
lamellar_stack_paracrystal
mass_fractal
mass_surface_fractal
rpa
sc_paracrystal
surface_fractal

[butlerpd]

A couple of comments/questions:

1. For models with "long range structure factor" does beta make any sense? in principle they cannot be multiplied by a separate structure factor because it is build into the model (e.g. binary hard sphere). That probably means no polydisperse parameters should be allowed? but would have to check.
2. For most of the models in the list of models without a separate F(Q) I'm not sure the beta approximation would be useful. For example I would guess it would be appropriate for pringle and maybe even star polymer, but not at all sure it is reasonable for perl_necklace or flexible_cylinder?