Simplified interface for computing I(q,dq)

[pkienzle]

The PR provides a calculator for computing I(q), I(qx, qy) and sesans G(ξ)-G(0).

To include support for single-point calculations a number of other pieces of code needed to change, including the resolution calculators and the SESANS calculator. These are on the SasView code path so need to be checked carefully.

The sasmodels internal data shim was modified to allow initialization with lists rather than numpy arrays. This does not affect SasView calculations.

The warning for invalid arcsin in SESANS was removed, and the mask updated to handle NaN values. This may affect the results of some SESANS calculations since the previous code did not exclude unreachable q values.

While debugging I noticed that the call to the calculator ignored misnamed model parameters. I modified the call to check that all parameters are used. This broke the hardsphere model tests since radius_effective is not a polydisperse parameter (PR#562). The tests will fail until that PR is pulled into master.

The call checks will not affect sasview, which goes through the SasviewModel code path to call the calculation kernel, but those using sasmodels directly with bumps may see errors popping up in their models if they have misnamed parameters.

[butlerpd]

need somebody with sesans interest/knowledge to take a look - note the detailed writeup referenced in #564 ... @rmdalgliesh?

[caitwolf]

I tested the infinite slit cases and the code worked as expected. It no longer ignored the slit resolution if only one of the length and width was provided.