Use spectral condensation to better-parametrize the LCFS

[dpanici]

Currently, a naive parametrization using the geometric poloidal angle as measured from the magnetic axis is used to parametrize the LCFS Fourier series for DESC from the EFIT R,Z points. This of course will not result in a condensed Fourier spectrum, and it is preferable to instead use spectral condensation to parametrize the boundary.

Ideally this would be implemented in DESC, but a standalone implementation here is probably faster first.

[dpanici]

Alternatively, can replace the geometric theta with just an arbitrary theta like theta = np.linspace(0,2pi, npts), may be better than geometric at least. check this