feat: add AI workflow snapshot command

[ProfRandom92]

Adds a minimal deterministic AI workflow snapshot command.

Validation:

• python -m compileall -q scripts/ai_workflow_snapshot.py
• pytest tests/test_ai_workflow_snapshot.py -q

[gemini-code-assist]

Code Review

This pull request introduces a new script, scripts/ai_workflow_snapshot.py, along with comprehensive unit tests to generate deterministic local AI workflow evidence snapshots. The script integrates with existing artifact and safety gate utilities to produce a compact JSON payload. Feedback was provided to enhance the robustness of the snapshot construction by strictly validating the validation_evidence field, ensuring it is treated as a list or raising a RuntimeError for unexpected types, which aligns with the repository's standards for processing JSON-like structures.

[gemini-code-assist]

To maintain strictness when processing JSON-like data structures, expected list fields should be treated as empty lists if missing or null, and a RuntimeError should be raised for other non-list types. This ensures robustness even if the underlying bundle structure changes, adhering to the repository's general rules for JSON processing.

Suggested change

	snapshot: dict[str, Any] = {
	"agent_artifact_bundle": agent_artifact_bundle,
	"ok": agent_artifact_bundle["ok"],
	"result": agent_artifact_bundle["result"],
	"safe_pr_gate": agent_artifact_bundle["safe_pr_gate"],
	"validation_evidence": agent_artifact_bundle["validation_evidence"],
	}
	validation_evidence = agent_artifact_bundle.get("validation_evidence")
	if validation_evidence is None:
	validation_evidence = []
	if not isinstance(validation_evidence, list):
	raise RuntimeError("validation_evidence must be a list")
	snapshot: dict[str, Any] = {
	"agent_artifact_bundle": agent_artifact_bundle,
	"ok": agent_artifact_bundle["ok"],
	"result": agent_artifact_bundle["result"],
	"safe_pr_gate": agent_artifact_bundle["safe_pr_gate"],
	"validation_evidence": validation_evidence,
	}

References

1. When processing JSON data, treat null or missing values for expected list fields as empty lists, but raise a RuntimeError for other non-list types to maintain strictness.