[trissim]

Six Official30 pipelines on main fail compilation. Repairing admission exposed failures in saved-image publication and native-image comparison. This PR fixes the determining owners and restores 30/30 native-reference comparisons, all with zero differences.

MeasurementSubject derives compatible row identity domains while preserving exact source-qualified image subjects and each payload owner's cardinality requirements. Compiled carrier admission follows authoritative predecessors through inherited bindings, retaining producer-transition and physical-source checks. Runtime anchors use the compiled main-input dependency, and execution groups use the declared payload owner rather than auxiliary filename inputs. Metadata targets enumerate actual declared writer destinations and reconcile durable typed projections after value cleanup; unknown image addresses still fail. Internal diagnostic planes declare terminal persistence without external export intent, shared persistence policies live in the materialization package, and source serialization preserves their nominal subtype.

Production source 4c4034835 includes main 2cbfc4a07 and its exact dependency pins, including PolyStore 84f322e. 586 local tests and 20 subtests pass, with one existing optional Napari skip. Original CI-pinned R0 and NRA R1 pass without increases; the original ClassDef census retains 5,122 declarations, 5,110 projected and all 12 OPEN rows. The final pushed documentation commit has an identical production/dependency/benchmark/script/test tree. Tests cover real multi-destination writes, cleanup and durable reopen, unknown-address rejection, implicit main-flow controls, stored cross-axis/direct-source execution domains and materialization-policy roundtrips. Three stale main fixtures were reproduced on unchanged main and repaired from their authoritative declarations/source fields and the existing shared ACK fixture; scientific assertions remain strict.

Native qualification progressed 22/30 → 29/30 → 30/30 on the synced production revision. Existing 1e-6 absolute/relative tolerances, exact discrete/identity checks and declared-image comparisons are unchanged. The production comparison suite reuses retained native scientific outputs; old native clocks/speedup columns are not fresh performance evidence. Ready-server startup/shutdown are excluded from pipeline clocks, and no tests/audits/builds/replays overlap timed qualification. No speedup or fresh scaling figure is claimed.

Validation, determining owners and reproduction and its authenticated evidence archive contain the original failure counterexamples, final native observations, generated pipeline sources, measured receipts, local test XML, source/dependency identities, original structural guards/census and NRA relocation evidence. NRA's configured syntax/import proof is separate from authored semantic behavior and native comparisons.

Performance grinding continues under #162. Shared generic runtime ownership remains separate in PR #326. The saved-image boundary extends #315 without claiming its separate parent/live acceptance is complete.

Fixes #330. Refs #315, #214, #324.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.