Allow structure factor models without volfraction or radius_effective

[pkienzle]

As noted in #719 (comment), the aggregation structure factors are different from those from liquid state theory and volume fraction is an irrelevant parameter.

For structure factor calculations, the P@S form checks that all structure factor models have radius_effective and volfraction parameters. This is because the structure factor calculation can transmit average radius of the particle to the radius_effective parameter of the structure factor (use radius_effective_mode != 0). Similarly, for hollow models like vesicle` the volume fraction of the enclosed volume is larger than the volume fraction of the shell. Computing the effective volume fraction also requires an average over the polydispersity.

We need to modify sasmodels/product.py to allow structure factor models with optional radius_effective and volfraction.

[pkienzle]

Strip unused volfraction and radius_effective from existing structure factor models. See #719.

[pkienzle]

volfraction currently uses the definition in P or the definition in S. If we move it from both then P@S will have no volfraction parameter. We could either invent one for the derived P@S model (making the structure factor code slightly cleaner), or continue to ignore it in the structure factor code, making the product model slightly cleaner. A third option is to drop volfraction from the model entirely and only use scale.