Find boundary psi without original

[bclyons12]

The find_psi_boundary function was updated to take the original_psi_boundary as an input. What should I do if I don't know the original boundary flux? In my case, I'm not starting from a dd and just want to find the boundary flux from a gridded psi.

[orso82]

IMAS.jl/src/physics/fluxsurfaces.jl

Lines 127 to 139 in f82999d

	# determine if this is a closed boundary equilibrium solution mapped to RZ grid like CHEASE would do.
	# When this happens the last closed surface touches the computation domain.
	# In this case we don't want to alter the original value of psi_boundary.
	psi_edge = [PSI[1, :]; PSI[end, :]; PSI[:, 1]; PSI[:, end]]
	if psi_axis < original_psi_boundary
	psi_edge0 = minimum(psi_edge)
	else
	psi_edge0 = maximum(psi_edge)
	end
	surface, _ = flux_surface(dimR, dimZ, PSI, RA, ZA, Float64[], Float64[], psi_edge0, :closed)
	if !isempty(surface) && ((abs((minimum(surface[1].z) - minimum(dimZ))) < 1E-3) || (abs((maximum(surface[1].z) - maximum(dimZ))) < 1E-3))
	return (last_closed=original_psi_boundary, first_open=nothing)
	end

[bclyons12]

Per discussion with @orso82, the routine currently assumes that if the minimum/maximum psi on the domain boundary corresponds to a closed flux surface, the user has probably provided a CHEASE equilibrium with concentric surfaces outside the plasma boundary. Thus, the routine returns the original boundary. This fails if a free-boundary solution is given where the domain boundary is the limiter. Commenting out the lines above fixes the problem.

We should think about how to fix/refactor this. Likely we want a flag for whether to perform this "CHEASE" check. Also, the original psi boundary may not be known, so it would be good if there was a way to specify concave vs convex. This is know from Ip and the COCOS as well.