Complete HOPS inputs retain embedded candidates, contaminants and censoring

Can native near/far-IR measurements support embedded-source representation without dropping unavailable targets?

Acquire the full five HOPS tables without class or score filtering; separate observables from references and preserve native zero-null and flag contracts.

Independent literal-byte/Decimal reader checks36956 values/nulls, all410 observable/reference joins, coordinates and four duplicate groups. Preserve unexecuted original reviewer and pre-execution flux-unit amendment.

410 targets,406 canonical sources:371 young-star candidates including93 Class0,29 likely extragalactic and6 uncertain. Six GETs total314416bytes; all terminals exit0.32 bindings; stage0.93MiB.

Next: Fit missing-view unsupervised SED factors from observables only, freeze before reference evaluation and compare reconstruction/raw/brightness/call-all baselines. Preserve embedded/star/cloud and fresh-survey validation.

Confirmed verdict covers inputs only. Target conditioning gives371/400 positive prevalence; call-all already exceeds90% precision. References remain provisional; no physical class, starless-cloud truth or observed birth adopted. Upper limits/non-detections are not measured zeros.