Add Agent Observability and Trace Analysis Workflow

Summary

Add an advanced observability workflow for agent runs, including traces, tool calls, retrieved context, latency, failures, and output analysis.

Proposed location

advance_ai_agents/agent_observability_trace_analysis/

What to build

Create a runnable example that instruments agent runs with OpenTelemetry-compatible tracing or local trace logs. It should capture prompts, tool calls, retrieval steps, latency, errors, model metadata where available, and final outputs.

Include a small analysis agent that reviews multiple traces and summarizes common failure patterns, slow steps, or tool-use problems. Phoenix by Arize, OpenTelemetry, or OpenLLMetry can be used where appropriate. Follow the repository's existing model-provider configuration pattern for model calls.

Acceptance criteria

• Includes a README.md based on .github/README_TEMPLATE.md.
• Captures traces or logs for agent runs and tool calls.
• Records latency, errors, retrieved context, and final outputs.
• Includes sample trace data or an easy local run to generate traces.
• Adds an analysis step that summarizes failure patterns across runs.
• Documents how to view or inspect the traces locally.
• Adds one catalog entry to the root README.md under Advanced AI Agents.

Contribution notes

Please keep this to one self-contained project and one pull request. Link the pull request with Closes #<issue-number>.