Implement simplified surfaces and faster contouring

[bclyons12]

This creates a simplified, lighter-weight version of flux surfaces suited to some applications (e.g., FRESCO). It also uses the new IMASutils contouring where applicable.

[bclyons12]

@orso82 I didn't modify flux_surface() since that has lots of fancy logic for open surfaces that I can't guarantee would be covered by IMASutil's contouring. It looks like we could do the same thing for trace_surfaces() that I did for trace_simple_surfaces(), and for the other version of find_psi_boundary() like I did for my version.

[bclyons12]

Also, once this PR is deemed ready to merge, this is a circumstance where we should run all the FUSE tests on it first. That will stress test the new contouring routines.

[orso82]

To run the FUSE tests using your changes, just open a PR in FUSE on a branch with the same simple_surface name that you have used here.

[bclyons12]

The IMAS 2.0 refactoring has been merged into this, which means both versions of find_psi_boundary() now use the new contouring.

[bclyons12]

@orso82 trace_surfaces() now uses the new IMASutils contouring as well. Only flux_surface (and other fucntions that depend on that) is left untouched because of the other types of surfaces that it can contour.

[bclyons12]

@orso82 All the FUSE regressions pass (ProjectTorreyPines/FUSE.jl#770) so I'll merge this unless you want to make the find_psi_boundary more robust under this PR.

[orso82]

@bclyons12 this is great, although it's unclear to me what's the difference between trace_surfaces and trace_simple_surfaces.

Do we really need both?

If we can't just have a single version, could you add a description to the docstring oftrace_simple_surfaces and perhaps highlight the difference with respect to trace_surfaces

[bclyons12]

@orso82 I cleaned up trace_simple_surfaces even more and added some documentation. The idea is that the FluxSurface struct has a lot of information, but if all we want is to perform flux surface averaging, we can get away with a lot less. In FRESCO where I might need to trace surfaces every iteration, this reduces the computation needed.

[orso82]

Thank you for doing this @bclyons12 ! 👍

I wonder if we should add this logic directly to the find_magnetic_axis() function.

[bclyons12]

You would have to pass the wall to it then, so up to you. You could make a separate function that takes the wall as an argument and does this I suppose.

[bclyons12]

@orso82 If the FUSE tests pass (ProjectTorreyPines/FUSE.jl#770), I would like to merge this.

[bclyons12]

All FUSE tests pass (ProjectTorreyPines/FUSE.jl#770)