Patching small issue with dr_sep for some equilibria

[kpentland]

Just patching a small issue that returned an incorrect value for dr_sep for some "weird" shaped equilibria. Should work fine for most but this should now work for all.

Main changes:

• instead of filtering the x-points by those that have psi values closest to psi_boundary, we instead keep only those inside the tokamak wall, then choose the two closest to the magnetic x-point.
• the values of psi at these two x-points are then used to find dr_sep.

[georgeholt1]

Thanks @kpentland! This method by filtering based on locations inside the wall is definitely an improvement. Couple points to discuss though:

1. The old way ordered the X-points lower-to-upper. In the docstring, it still says: "By convention, the function assumes the first X-point corresponds to the lower X-point, and the second to the upper X-point.". With this change, this is no longer necessarily the case. I can't think of a reason why we would need to maintain this convention, so my suggestion is to either fix the docstring or do a sort on the X-points:

   closest_xpts_sorted = closest_xpts_sorted[np.argsort(closest_xpts_sorted[:, 1])]

2. _updatePlasmaPsi uses a local opt variable and only sets self.psi_axis = opt[0][2]. The new code references self.opt[0, 0:2], but this could potentially become not up-to-date. I suggest either using self.magneticAxis()[0:2], or update _updatePlasmaPsi to that self.opt is reliably set.

3. Obviously calling this function for single-null plasmas doesn't make sense, but it might still happen if people are using it in an automated workflow. Therefore, it would be an improvement to have a warning so that the downstream closest_xpts_sorted[1, 2] doesn't raise an index error:

   if xpts_inside_wall.shape[0] < 2:
       warnings.warn("Fewer than 2 X-points found inside the wall; cannot compute dr_sep.")
       return [None, None]

4. # get indices of the two cloest to magnetic axis
   cloest --> closest

[kpentland]

Hi @georgeholt1,

Thanks for this!

I've made the changes so that the code reflects the docstrings and purposely uses the lower X-point first (I think this is a typical convention and if different for other tokamaks, people can flip the signs). It also now throws an error if less than two X-points are found within the first wall. I've also called the eq.magneticAxis() function to make things more stable.

[georgeholt1]

Thanks again @kpentland