Retabulate capture-efficiency hydro columns on the hinge basis

[Copilot]

The capture-efficiency CSVs mixed hinge-measured P_capture_W with CG-referenced diagnostic hydro columns, which made masked basis-wrong and allowed negative de-normalized B55 values in Python. This updates post-processing to use the hinge-referenced H5s consistently, match the C++ non-negativity clamp, and add a no-resim retabulation path for existing CSVs.

• Basis correction in sweep post-processing

  • Switched passive_vs_optpassive_sweep.py, capture_efficiency_sweep.py, and cc_capture_efficiency_sweep.py from hydroData/vgoswec_*.h5 to hydroData/hinged_vgoswec_*.h5.
  • This puts B55_Nmsrad, F_exc_Nm, P_opt_W, masked, and derived eta on the same hinge basis as P_capture_W.

• Python/C++ de-normalization parity

  • Clamped Python de-normalized damping to non-negative values before masking / P_opt evaluation, matching src/impedance.cpp.
  • This removes nonphysical negative B55 from the tabulation path without changing the existing mask rule.

b55 = np.maximum(0.0, b55_norm * rho * w)
masked = b55_t <= MASK_B55_THRESHOLD

• No-resimulation re-tabulation entry point

  • Added scripts/retabulate_hydro_columns.py.
  • Reads existing analysis/{passive,opt_passive}/capture_efficiency_VGM*.csv, preserves P_capture_W and provenance fields byte-for-byte, recomputes only hydro-derived columns, and writes back with existing formatting conventions.
  • Supports --dry-run with per-flap masked-row change summaries.

• Stage plotting from committed CSVs

  • Added scripts/plot_opt_passive_stage.py.
  • Writes:
    • docs/img/opt_passive_stage_per_flap.png
    • docs/img/opt_passive_stage_all_flaps.png
    • docs/img/opt_passive_stage_gain.png
  • Uses docs/freedecay_validation.csv (cpp_zerocross_wn_rads) for free-decay T_res markers and validates paired period grids before plotting ratios.

• Docs + regression coverage

  • Updated analysis/README.md and analysis/BENCHMARK_NOTES.md to describe hinge-basis tabulation and per-file rho/g reads (rho=1025 in hinged files vs prior CG-basis note of 1000).
  • Updated docs/EOD_SUMMARY_2026-09-19.md to mark the hydro-tabulation follow-up as resolved and keep the isolated resonance-adjacent P_capture dips explicitly open.
  • Added regression coverage for:
    • non-negative Python B55
    • hinge-vs-CG mask divergence
    • retabulation preserving P_capture_W