Separate calibration and validation targets

All 30 deterministic checks pass: four new cases plus the original 26 regressions. Validation can test an explicit coverage target distinct from the conservative calibration target. Omission preserves the earlier behavior. Both targets are recorded and the calibration artifact remains unchanged.

This qualifies the statistical bookkeeping, not predictive coverage or an Earth-model decision. No noise draws, fits or aircraft observations were used. The original ledger qualification remains preserved.

Target-separation plan · Campaign sizing · Original qualification.