feat(agent): add local operations agent

[xsuler]

Summary

• add areno agent as a local operations assistant for train and serve workflows
• add explicit user-input and user-response tools for interactive clarification
• stream command output through a PTY so long-running commands update live and preserve terminal colors
• document setup and usage in README and CLI docs

Motivation

AReno train and serve commands often require environment-specific choices around GPU memory, tensor parallelism, rollout concurrency, dataset loaders, reward functions, and recovery from setup failures. The operations agent gives users a guided local workflow without hiding the underlying commands.

Closes #156

[gemini-code-assist]

Code Review

This pull request introduces a local operations agent CLI (areno agent) and reusable coding-agent loops/tools to assist with AReno training and serving tasks. It also changes the default model hub from Hugging Face to ModelScope across the CLI, updates documentation, and adds smoke check options (--smoke-infer and --smoke-train) to the training CLI. The review feedback highlights several critical improvements: implementing retry logic for batch chat completions to prevent transient failures from crashing the entire run, preventing workspace resource leaks during batch initialization, securing sensitive configuration file creation to avoid permission race conditions, and catching Exception instead of BaseException to preserve proper interrupt handling.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.