Retain canonical Swift source-sink qualification evidence

[DavidBakerEffendi]

Description

The Foundation corrections had only synthetic diagnostic controls. Preregister a bounded runner and retain fresh observations over the unchanged canonical Swift source–sink pair, with original metadata/population joins and portable source archives. Refs #220 and #215; both issues remain open and acceptance is incomplete.

Key Changes:

• Under prospective commit 7cdce9ad, the adapter-assisted lane observes one positive source-to-command flow and zero constant-command negative flows, with one recognized source and two sink arguments in each case. The separate vendor-native lane observes zero sources/flows and two sink arguments in both cases; silence is not converted into unsupported.
• Bind runtime/configuration/runner identities, retain successful extraction/query/decode and raw logs, and add portable CI verification plus 14 focused tests. Extraction took 89.54/87.58 seconds within the 150-second cutoff; queries use 60 seconds/2048 MiB. Aggregate memory and semantic completeness remain unproven, activation stays pending, and all evidence is non-scored.

Touch Points:

• adapters/codeql/swift-native-v3/qualification-plan.json
• scripts/run-swift-native-qualification.py
• scripts/verify-swift-native-qualification.py
• scripts/test-swift-native-qualification.py
• evidence/swift-native-qualification-v1/
• docs/swift-canonical-qualification.md
• docs/src/content/docs/swift.md
• .gitattributes and .github/workflows/ci.yml

Validation: canonical verifier and all 14 tests passed from a clean Git archive; pending integration verifier/tests, retained Foundation/Joern verifiers, v2 population audit, cargo fmt --check, and git diff --check pass. No fixture binaries ran. Historical reports, freezes, population bytes, opaque decisions and Joern's blocker remain unchanged.