The prediction ranges did not improve
The study reached its 40-minute cooperative ceiling. It preserves 39 passing sector calculations, four interrupted starts and five unstarted sectors. No process was killed and no allowance was extended. The study is closed; its unfinished work will not be retried under this allowance.
All four globe candidate/processing groups completed their eight-sector comparisons. Their combined prediction ranges are exactly the same as the inherited ranges. Both specified-field groups still miss the fixed width target. Completing the remaining disc sectors cannot change that failed all-group gate. The wider fresh ranges remain visible in the complete report.
The saved baseline audit explains the main limitation. In all 39 completed cases, the existing training-derived acceleration bound is tighter than the new physical-only bound in every component and in vector norm. The promising static comparison used a broader baseline than the prediction method actually used. That comparison was insufficient grounds to expect a complete improvement.
The static and native controls remain valid component results. The next method must preserve the training information and address direction loss in dynamic propagation. No aircraft fit, measurement or published 53-stretch finding changed.