chi is written without the signgs factor carried by phi, phipf and chipf

[CharlesCNorton]

eqfor.f90 integrates chi without signgs:

! eqfor.f90:158-163
  chi1(1) = zero
  DO i = 2, ns
     ...
     chi1(i) = chi1(i-1) + hs*(phip(i)*iotas(i))
  END DO
  chi = twopi*chi1

The quantities it has to agree with all carry it: phi = (signgs*twopi*hs)*phi (fileout.f90:61), and phipf and chipf go into the wout as twopi*signgs*phipf and twopi*signgs*chipf (wrout.f90:736, 738). chi goes in as is (wrout.f90:737).

With signgs = -1 the file holds chi = -∫ iota dphi. On input.cth_like_fixed_bdy:

quantity	value
phi(ns)	-0.035000
Σ iotas(js) phipf(js) hs over the half grid	-0.040086
trapezoid of chipf over s	-0.03846 (the axis entry is zero, see #6)
chi(ns)	+0.040086

chipf(js)/phipf(js) equals iotaf(js) on every surface, so chi is the odd one out, and d(chi)/d(phi) read back from the file has the opposite sign to iotaf. PARVMEC carries the same three lines (eqfor.f:177-183), so this is inherited from 8.52. chi = twopi*signgs*chi1, or accumulating phip the way fileout.f90:61 does, fixes it.