InnerLayer.SLAYER - Build P_perp from D_perp by default, not chi_perp_e #508

Summary

The InnerLayer (IL) SLAYER model builds the perpendicular Prandtl number from the electron perpendicular heat diffusivity, P_⊥ = τ_R χ⊥,e / r_s² (src/InnerLayer/SLAYER/LayerParameters.jl, tau_perp = rs^2 / chi_perp; chi_perp is fed from the kinetic-file chi_e profile in src/Tearing/Runner/run_slayer.jl). In the Fitzpatrick layer formulation P_⊥ = τ_R / τ_⊥ is set by the perpendicular diffusivity D⊥, which combines c_β, η, χ⊥,e, χ⊥,i and the n, T gradients (Fitzpatrick, Phys. Plasmas 29, 032507 (2022), Eq. 16). χ⊥,e alone is not that quantity, so every SLAYER growth rate and frequency is computed with the wrong P_⊥.

Proposed change

#501 adds [SLAYER] P_perp_model and implements the D_perp closure, but keeps chi_perp_e as the default so that it moves no results. Once #501 merges:

• make P_perp_model = "D_perp" the default;
• keep chi_perp_e selectable for comparison with earlier results, documented as not the physical P_⊥;
• keep chi_perp_e setting the critical-Δ dc_tmp, which is unaffected.

Who is affected

Every SLAYER run that does not set P_perp_model explicitly. SLAYER P_perp, Q_root, ω, and γ will move, including the diiid_slayer_n1 regression case, so the PR carries !. The D_perp closure needs a chi_i profile, which #501 adds to the kinetic HDF5 schema and to the DIII-D-like example kinetic file; the fallback when chi_i is absent should be decided in the PR.