Take the vacuum magnetic pressure alone as the total pressure outside the free boundary

[CharlesCNorton]

funct3d.f90 adds the extrapolated edge pressure to the vacuum magnetic pressure before using it as the total pressure outside the plasma:

presf_ns = pmass(hs*(ns-1.5_dp))
IF (presf_ns .ne. zero) then
   presf_ns = (pmass(1._dp)/presf_ns) * pres(ns)
end if
...
gcon(l)      = bsqvac(lk) + presf_ns

bcovar has already added the kinetic pressure to bsq, so the plasma side bsqsav(lk,3) carries p + |B|^2/2mu0. Adding presf_ns on the vacuum side cancels that pressure from both sides, so rbsq and dbsq measure continuity of the field strength instead of the pressure balance p + |B|^2/2mu0 = |B_vac|^2/2mu0 that holds across a plasma-vacuum interface. This takes the vacuum magnetic pressure alone and drops the extrapolation, which nothing else reads.

The same change is ORNL-Fusion/PARVMEC#86 and proximafusion/vmecpp#830.

presf_ns is identically zero for a pressure profile that vanishes at s = 1, so every input that ships is unaffected and ctest passes.

It matters for a profile with a pedestal. Starting from fb/input.cth_like_free_bdy with pmass_type = 'power_series' and AM = 1.0, -0.5:

edge pressure	boundary, max	boundary, rms	volume	iota on axis
20 Pa	0.556 mm	0.260 mm	0.304334 -> 0.304593	1.255122 -> 1.255816
100 Pa	2.854 mm	1.320 mm	0.304827 -> 0.306134	1.256954 -> 1.260467

linear in the edge pressure, and within one percent of what the same change gives in PARVMEC on its own copy of that case.