Zeff profile support

[StuartBenjamin]

Lets RDCON's modified Rutherford equation (MRE) terms use a radial Zeff profile. Before, mufrac used one scalar Zeff on every surface.

Contains #295 (vacuum matrix in the plasma's Fourier frame), which should be merged first.

Changes

• &RDCON_CONTROL takes Zeff and psi_N_Zeff as arrays of up to 999 points. The new read_var_len (equil/local.f) copies the given entries into allocatable arrays.
• With MRE_flag, mercier.f splines Zeff(ψ_N) and evaluates it on each surface for mufrac.
• Without psi_N_Zeff, a single Zeff value, or no Zeff (then 1.52, the old default), gives a flat profile, so existing inputs run as before.
• Zeff and psi_N_Zeff must have equal lengths and at least 4 points (the "extrap" spline end condition). Shorter tables used to give NaN and now stop with a message.
• The rdcon netCDF stores Zeff and psi_N_Zeff as global attributes when MRE_flag is set. Without MRE_flag it no longer writes a Zeff attribute.
• docs/examples/DIIID_resistive_example/rdcon.in sets MRE_flag and geom_flag to f.

Tests

RDCON with MRE_flag = t on a 513-surface TokaMaker i-file, compared with develop:

rdcon.in	Result
no Zeff	mufrac, Dnc match develop to 3e-9 (develop's default 1.52 was a single-precision literal)
Zeff = 2.0	match develop to 1e-16
Zeff = 2.0 ×4, psi_N_Zeff = 0, 0.3, 0.7, 1	same as Zeff = 2.0 to 3e-16
2-point table	stops with "Zeff, psi_N_Zeff: equal sizes, >= 4" (was NaN)

The GPEC build and the install tests pass, apart from the known a5_tearing STRIDE failure, which this PR does not cause.

---

🤖 Generated with Claude Code

[StuartBenjamin]

New commit 6eda483e: MRE runs without a Zeff table use a flat Zeff, as before.

With MRE_flag = t, an rdcon.in that gives no psi_N_Zeff used to stop with "Provide Zeff profile values". On develop the same input ran with the scalar Zeff (default 1.52), so older input files broke.

• One Zeff value without psi_N_Zeff, or no Zeff at all (then 1.52), now gives a flat profile at that value.
• The profile spline's "extrap" end condition reads 4 nodes. Tables with fewer points used to give NaN in mufrac and Dnc without any error, so they now stop with "Zeff, psi_N_Zeff: equal sizes, >= 4". The separate pchip PR fits the Zeff table with pchip, which handles 2 or more points, and relaxes this check to 2.

Tests (RDCON on a 513-surface TokaMaker i-file, MRE_flag = t):

rdcon.in	Result
no Zeff	runs; mufrac, Dnc match develop to 1e-9 (develop's default 1.52 was a single-precision literal)
Zeff = 2.0	runs; mufrac, Dnc match develop to 1e-16
Zeff = 2.0 ×4, psi_N_Zeff = 0, 0.3, 0.7, 1	identical to Zeff = 2.0
2-point table	stops with the message above (was NaN)

---

🤖 Generated with Claude Code

[matt-pharr]

Looks good, I made the few tweaks here that are relevant.