⏯️ [FEATURE] [1.0] Workflow Lifecycle Controls

Is your feature request related to a problem?

Once a workflow is running there is no official way to pause, resume, or cancel it via REST. Operators need to intervene when downstream systems detect problems or when long-running cascades should be aborted. Manual intervention (e.g., restarting the service) is risky and imprecise.

Describe the solution you want to see

• Provide lifecycle endpoints (e.g., POST /api/v1/workflows/{correlation_id}/cancel, /pause, /resume) that manipulate pending tasks for a given correlation ID.
• Surface status transitions and reasons (cancelled by user, paused for approval, auto-resumed) in the status endpoint and traces.
• Respect visibility/auth constraints so only authorized roles can control workflows.
• Ensure cancellation propagates cleanly: running agents are interrupted where possible, pending tasks are dropped, and affected artifacts are marked accordingly.

Describe alternatives you have considered

Today operators rely on ad-hoc flags or manual data edits, which are error-prone and lack audit trails.

Additional context

Target for 1.0 so it launches alongside advanced auth (#280) and persistence enhancements. Lifecycle control should integrate with human-in-the-loop and approval flows to give a full operational story.

The issue names no files, tests, or entry points; begin by locating the existing REST workflow and status endpoints, task persistence, and trace handling. Review the authentication work in #280 and the human-in-the-loop and approval paths. Done requires an agreed lifecycle design covering authorization, status and audit details, cancellation propagation, and affected artifacts.