Frozen dust models verified; stellar shortlist lacks nearby cached clumps

Can the existing stellar shortlist support compatible dust-cloud assessment?

Replay five spatial E29 dust PCA/probe models and retain every cloud alternative at fixed3/15/30arcsec, plus60arcsec declination controls; preserve all stellar flags.

Independent native feature/mask/scaling/PCA review compares1,892,080 latent coordinates,94,604 held-out scores and94,604 OOD states. Exhaustive separate geometry checks600,735,400 candidate/cloud admissions in both arms, all3,175 records and counts.

All94,604 frozen dust predictions reproduce, max probe difference4.44e-16. Zero real or control cloud alternatives at every radius; all3,175 candidates remain explicit no-nearby-cached-clump cases. No new fit, catalogue acquisition or duplicated cloud arrays.

Next: Screen additional unfiltered plane fields that intersect cached clouds, retaining all alternatives and independent ordinary/young/contaminant reference coverage; calibrate before fresh validation and perform astrometric/temporal refinement.

Zero proximity is not physical cloud-free/starless evidence. Current shortlist and cached subset do not overlap at these radii. Unsupervised PCA but supervised catalogue-evolution probe; same-input raw comparator remains better and its coefficients unavailable for replay. No independent truth, ownership, membership or confirmed birth.