Add GET /api/workflows/{workflow_id}/metrics endpoint for workflow analytics

[akgohain]

Motivation

• Provide a backend endpoint to surface workflow-stage and agent-action analytics as a stable API for monitoring and UI consumption.
• Ensure the endpoint returns safe defaults when a workflow has sparse or no events and produce a stable response schema for downstream consumers.

Description

• Add a new workflow router at server_api/workflow/metrics.py that implements GET /api/workflows/{workflow_id}/metrics and a _compute_workflow_metrics helper to derive analytics from an in-memory workflow_events store.
• The metrics computed include event counts by type, approval/rejection counts and rates, stage transition counts formatted as from->to, and first/last event timestamps (ISO8601), with safe defaults when events are absent.
• Wire the new router into the app by including it in server_api/main.py and add focused unit tests in tests/test_workflow_metrics.py covering empty workflows and seeded event scenarios.
• An example API schema (returned JSON) is included with the change and follows the shape: {"workflow_id","event_total","metrics": {"event_counts", "approvals", "stage_transitions", "timeline"}}.

Testing

• Ran uv run pytest -q tests/test_workflow_metrics.py, which failed in this environment due to an editable dependency metadata issue for pytorch_connectomics (package metadata generation error).
• Ran PYTHONPATH=. pytest -q tests/test_workflow_metrics.py, which executed the new tests successfully and reported 2 passed.

---

Codex Task

[akgohain]

Consolidated into draft PR #165 (feat/codex-wave-integration). Closing this source draft to keep review in one place.