Add ne_greenwald_fraction to PostProcessedOutputs

[theo-brown]

The Greenwald density fraction is generally useful. Having it in the simulation state is useful as I am interested in developing feedback controllers for density fuelling, with the Greenwald fraction as the target variable (eg #532).

[jcitrin]

Thanks for this.

Can have a short discussion on appropriateness of line-averaged-density vs volume-averaged-density.

In the literature I think line-averaged-density is more often used. For a given machine it's easier to measure (no need for full geometry reconstruction) but at the same time the specific chord used is very machine specific.

Volume-averaged-density is less ambiguous.

However, if you are developing controllers for density feedback, then should think about what the appropriate synthetic diagnostic is. Is the fGW provided by the real diagnostic a line average with a specific chord, or volume averaged following geometry reconstruction? Likely the former, and then you'd probably want a more realistic synthetic diagnostic in TORAX with the actual chord, that would then necessitate more 2D geometry information propagated in TORAX etc etc.

For now happy to approve this volume averaged one, but wanted to raise the above points. Feel free to include some of these points in the docstring of the method if relevant. Am also wondering if the name should change to e.g. ne_volume_avg_fgw, or fgw_ne_volume_avg, or something like that, to accentuate the density used for the fraction.