new R_eff branch of sasmodels does not support VR

[pkienzle]

The new effective radius infrastructure does not support VR functions. There are 4 models that need VR: vesicle, hollow cylinder, hollow rectangular prism and hollow rectangular prism thin walls.

This should use the same math as SasView? 3.x, scaling the structure factor volume fraction by the average particle volume / average shell volume.

See also Ticket #1030.

To implement this, we need to touch the same files that we touched for effective radius on the new R_eff branch:

modelinfo.py, generate.py, product.py kernel*.py, kernel_iq.c

We are already computing the shell volume in the form_volume function.

We could rename this to shell_volume for those functions that need it, and define form_volume as the total volume. This will make it easier for future maintainers to figure out what is going on. We can define some macros in generate.py/kernel_iq.c which return shell_volume if it is available, or form_volume as shell_volume.

For P(q) return F2 / shell_volume. For the call to S(q) set the volume fraction to volfraction * form_volume/shell_volume.

[pkienzle]

The volfraction name is confusing. Consider renaming the parameter to volfraction_shell in vesicle, and in the product model if shell_volume is available in the model. This could still be misinterpreted as the ratio of shell to particle rather than volume fraction of the shell in the entire solution, so better names would be nice.

[pkienzle]

See #1202.