[cmccomb]

Change

Add independent accumulated PTO-work gates to three existing MATLAB comparisons. No production solver, physical default, or badge count changes.

• Published Sphere Variable_Hydro/Variable_Mass (900 s): gate PTO speed at 0.005 m/s, pointwise absorbed power at 2.5 kW, and integrated work at 1%. Observed maxima are 0.004135 m/s, 1.052 kW, and 0.00376% work.
• Configured RM3 spring/pretension variant (400 s): reconstruct irreversible damper power from MATLAB's saved PTO speed, separate from spring exchange. Ramp, two steady, and full-run work windows pass the preset 1% limit; the largest observed gap is 0.7934%. This variant is not an additional published input.
• Published OSWEC Multiple_Wave_Spectra (100 s): gate absorbed work in three disjoint intervals and the full run at a preset 5% limit. The largest observed gap is 3.913%.

The MATLAB traces are pinned fixtures. Each Python run advances independently from its published hydrodynamic and wave inputs. PARITY.md and fixture records contain the measured limits; the 52/63 strict count is unchanged.

Verification

8 passed locally across these three comparisons and the stacked RM3 End Stops gate. tools/render_parity_badge.py --check and git diff --check pass.

This PR is stacked on #603; retarget it to master once #603 merges.