Add an extra (optional) point at the end of the perturbation interval

[janmalec]

The perturbation procedure from Sandy sometimes produces interpolation artifacts when the energy grid of the initial cross section is too sparse. This can be seen on the attached images, where the lines labeled "Broadr + perturbed" and "Broadr + perturbed + proc" feature a smooth tail from the maxiumum value to zero at the edge of the perturbation. From the code sample from the custom perturbation function:

            enew = np.union1d(self.data.index.values, pert.right.index.values)
            u_xs = self.reshape(enew)
            u_pert = pert.reshape(enew)
            u_xs.data[(mat, mt)] = u_xs.data[(mat, mt)] * u_pert.right.values

it is clear that the point marking the edge of the perturbation interval is added to the total grid, but another point right next to it could be added to define the edge of the perturbation. This additional point can be optional and selected by the user. I am currently working on this feature and will submit a pull request soon.

The same images show another potential issue in the sampling procedure, where the cross section can change because broadening happens after the perturbation, but I will open a separate issue for that and submit a separate pull request (edit: reference: #254 ) .

[luca-fiorito-11]

See #254