Verify authorized Swift Objective-C opaque model activation

[DavidBakerEffendi]

Exercise the authorized Darwin/Foundation/Objective-C construction for the two opaque propagator templates with runtime fidelity and independent model-off/on controls. The declared model is load-bearing and respects the second-argument binding.

Key Changes:

• Retain preregistered selector-owner/signature/return checks and identical runtime/analyzer dispatch-class bytes. Both carry and block propagate at runtime; block stays undeclared in the model. With the model off only the direct baseline flows; model-on additionally reaches carry and argument 1, while the sibling, argument 0 and clean input remain absent.
• Preserve the initial compiler failure and inconclusive top-level analyzer attempt. Typed diagnosis proves those calls resolve but lack data-flow nodes; the separately preregistered function-scoped control keeps the dispatch body and queries unchanged. Add 16 portable runtime/activation tests to CI.

Touch Points:

• evidence/swift-opaque-objc-v1/
• scripts/*swift-opaque*.py
• docs/swift-opaque-objc-prototype.md
• .github/workflows/ci.yml

Validation: 7 runtime tests and 9 activation/provenance tests, cargo fmt --check, and git diff --check pass. The successful function-scoped extraction took 94.530s; all three queries and decodes completed under the 150s extraction / 60s query / 2048 MiB analysis request. Runtime selector evidence is distinct from static wrapper binding; no resolved dynamic perform target is claimed.

Non-scored prototype only: no population registration, other-analyzer qualification, whole-profile qualification, freeze or publication. Aggregate memory compliance and semantic completeness remain unproven.

Refs #220 and #215. Both issues remain open; acceptance is incomplete.