Separate assessment of the nine remaining stretches

Nine of the twelve unfinished comparisons were excluded from the cost-only recovery because their stopping failures included stationarity or predicted position stability. This study keeps them separate from the three new budget failures and preserves their earlier outcomes.

Every one of these nine receives all three physical models under both processing assumptions, starting from its frozen primary endpoints. Each receives the original predetermined perturbed-start repeat, and two completed recordings provide fresh controls. The qualified wider arithmetic and unchanged scientific assumptions are used throughout. There is no selection by favorable costs or gyro values.

The finite plan permits 65 starts and 12,810 native sweeps: 54 primary fits and nine repeats at up to 200 fitting evaluations each, two controls at forty each, plus two diagnostic predictions per start. One worker uses four numerical threads. There are no retries, relaxed stopping gates or automatic budget increases.

This stage launches only after the preceding budget continuation finishes with passing controls. Its target list and allowances are frozen before that preceding stage runs; outcomes cannot change which stretches are included. Each complete comparison still requires six passing primaries, its repeat and both fresh controls. Unfinished comparisons remain unknown. No global-optimum proof or calibrated scientific significance follows from numerical completion.