Towards a Playground to Democratize Experimentation and Benchmarking of AI Agents for Network Troubleshooting