E01 repaired-CAD matching supplement; all earlier evidence retained

Latest composite: results-volume-v2/E01_summary.json and E01_web.json.
Latest figures: figures-volume-v2/E01_driver_response.png/.svg and
figures-volume-v2/E01_captive_impact.png/.svg.

The original40-run matrix remains in results/. A CAD check subsequently found
four bosses tangent to, rather than overlapping, the enclosure wall. Increasing
their radius6to7mm repaired physical connectivity. The net air-volume proxy
changed from0.000706091410953079 to0.0007039780004475063 m^3, a0.2993% decrease.
The original inputs/results remain evidence for the earlier geometry; they
are not represented as matching the repaired assembly.

E01-VOLUME-REVISION-PLAN.json froze this geometry correction before its runs.
Exactly486 ND64 frequency rows and12 ND64 ODE integrations were recomputed.
The current972-row/40-integration view inherits486 ND91 algebraic rows and
28 unchanged integrations:12 ND91 plus16 impact. Historical matrix count is
40 original +12 correction =52 integrations. No impact physics was rerun.
The17 author tests separately included5 short test integrations. Zero physical
trials, built device validations or new CAD-dependent force laws are implied.

Runtime for the targeted correction:36.09400000004098 seconds. The original
matrix took137.10999999998603 seconds. Every source/parent artifact checked by
the correction retained its hash. All original and corrected numerical screens
and the preserved soft-capture failure remain explicit. Independent reviewer
packages are separate from this numerical source/trace allowlist.

From operating-models/, after installing Python, NumPy and SciPy:
  python -m unittest test_operating_models -v
  python publication-v2/verify_package.py --require-traces
  python run_volume_revision.py --output results-volume-rerun-001
  python make_figures_v2.py --results results-volume-rerun-001 --output figures-volume-rerun-001

The targeted runner inherits the frozen original results/ tree. Download the
separate NPZ traces to the exact relative paths in TRACE-INDEX.json for full
parent-evidence verification. The core ZIP alone can run author tests and
inspect all JSON results; it cannot verify omitted trace payloads. To verify
only core files, omit --require-traces; the checker reports that partial scope.
Full matrices can be independently recalculated with run_study.py in a fresh
output directory. Do not replace original results/ with such a rerun: timestamps
and entire summary hashes differ and would fail the correction's frozen-parent
hash check. Choose a new output path for every new calculation.

FIGURE HISTORY
figures/ is the original, superseded1ms force-curve draft.
figures-v2/ fixes peak-force presentation but retains the historical volume.
figures-volume-v2/ is the final repaired-volume figure set. Force bars use the
reported every5us-step peaks; motion curves are explicitly downsampled.

FIELD CLARIFICATION
In corrected per-ODE A_ND64...volume_v2.json details, evidence_origin copied
the configuration label 'new_corrected_volume_algebra'. That field describes
the copied configuration context and is mislabeled for the ODE record. The
authoritative acoustic_reports entry in the composite summary correctly says
'new_corrected_volume_integration'. The twelve detailed ODE traces, durations,
RHS counts and source hashes establish that they were actually integrated.
This correction note preserves the original detailed records unchanged.

SAMPLING FINDING
Read publication/REVIEW-NOTES.txt and the independent review companions.
Some acoustic dense-output cumulative-energy samples have tiny decreases;
exact monotonicity is not claimed. No result was retuned to remove this finding.
The corrected coarse ND64/CAD/20Hz case also fails the original author's
nonnegative-ledger gate: returned energy-3.1023928854649686e-12 J at12.5ms,
below-1e-12 J. Its fine counterpart passes at-3.551261472306674e-14 J.
Thus composite acoustic_numerical_checks_passed is FALSE, even though all
refinement comparisons pass. LATEST-EDITION.txt and diagnostics/ retain this
failure and the separately frozen post-hoc closed-form/quadrature check.
The nominal linear model has no measured SPL, acoustic focus, thermal protection,
fracture, powered reset/BPM or certified physical capture outcome.
