Sharper column bounds do not resolve the remaining width
All 48 sectors passed correctness and their independent endpoint checks. Complete widths improved by at most 0.016% relative to the preceding orientation-domain comparison. Height widths were essentially unchanged. The same three of six groups still miss the unchanged twofold target.
The study completed in 122.3 seconds, with zero native integration, new bounds, original aircraft reads or observed fits. All physical limits, prior constraints, original sectors and 51 coordinates remain. Applicable known truth stayed feasible and contained after predictions were saved.
More rotation-matrix inequalities are now a low priority. They can reject impossible examples without materially changing the complete ranges. The next diagnostic will isolate groups of internal error terms in saved generated cases, to identify where the remaining conservatism comes from before another refinement.
That diagnostic may temporarily supply known simulated values. Such a result would not be an available prediction method or a valid aircraft uncertainty interval. The 53-stretch findings and requirements for longer predictions are unchanged.