Bug in OML-current for kappa=3 and alpha=0.

[diakodarian]

Tasks

1. Proper handling of invalid kappa and alpha values in OML-current calculations.
2. Add tests covering invalid values for various input parameters.

Closes #34

[sigvaldm]

The code quality is good, but why the stronger condition on kappa when alpha is non-zero? I don't see any such clause in the paper, and the following seems to work on the master branch (though I haven't checked that the numbers actually makes sense):

In [8]: OML_current(Sphere(r=1e-3), Electron(kappa=1.5001, alpha=0.2), eta=10)
Out[8]: -8.76673817729523e-06

In [9]: OML_current(Sphere(r=1e-3), Electron(kappa=1.5001, alpha=0), eta=10)
Out[9]: -8.764108453694125e-06

One of the tests on finite_radius_current have also started failing because it gets thrown a ValueError, but that can be dealt with afterwards. (I support your usage of ValueError).