Tracking: OpenFUSIONToolkit findings to report upstream

[d-burg]

Three OFT-level behaviors discovered during the v1.2.0 work (measured on OFT base v26.6, branch fix/bootstrap-pchip-derivatives @ 19e08b2, DIII-D mesh). All are guarded/routed-around inside bouquet, but they deserve upstream reports so other OFT users don't rediscover them:

1. compute_flux_integral is not a plasma-domain integral. It integrates over the entire reg == 1 limiter region, and gs_prof_interp_apply CASE(4) returns 0 off-plasma so gs_flux_int evaluates the flux function at its LCFS value there. Measured: FI(1) = 2.83853 m^2 (limiter area) vs true plasma cross-section 1.79005 m^2; a current-density profile with finite edge value is over-counted by ~j_phi(1) x 1.05 m^2 = +11.9% of Ip on this equilibrium. bouquet's mitigation: utils.Ip_fsa_integral (v1.2.0). Also see Standard l_i loop's Ip root absorbs a +16-17% inductive-amplitude bias (compute_flux_integral is not a plasma-current measure) #15.

2. get_q(psi=...) silently collapses onto the magnetic axis if the requested grid contains psi_N = 0. All returned ravgs become surface-independent (<R> constant to 2e-15 across 257 surfaces), no exception raised, downstream results ~37% wrong. bouquet's mitigation: grid clipped to [1e-3, 1-1e-3] with a collapse assertion (v1.2.0, fsa_current_geometry). An upstream fix (raise, or clamp internally) would protect all callers. Related symptom seen at scan time: gs_get_qprof: Trace did not complete at every surface.

3. get_q ravgs dV/dPsi is per dimensional psi, undocumented. ∫ dV/dPsi dpsi reproduces get_stats['vol'] to -0.25%; the dpsi_N reading is off by +291%. The needed Jacobian is |psi_bounds[1]-psi_bounds[0]|. A docstring note upstream would suffice. (Also: sauter_fc exposes <R>, <1/R>, <a>, <|B|>, <B^2> but no V', so get_q dict-form ravgs — OFT >= 26.6 — is the only V' source.)

Suggested route: C. Hansen is a co-author on the bouquet paper; a direct note or OFT issues citing the numbers above. Validation details and test pins live in bouquet v1.2.0 (tests/test_fsa_current_integral.py).