Add Wave Terminal bridge and setup docs

[mizoz]

Summary

• add a minimal wave-bridge.js entry point backed by codex exec
• add npm run wave:server
• document a Wave Terminal setup flow using a local OpenAI-compatible endpoint

Why

Wave Terminal already supports custom OpenAI-compatible chat endpoints, but there is no native Codex OAuth mode. This adds a simple local bridge path that works well for Wave and similar clients, and also serves as a fallback when the SDK-backed server does not start cleanly in a given environment.

Notes

• no secrets or local config files are included
• the CLI-backed bridge is intentionally minimal and does not yet preserve multi-turn session state between requests
• the existing SDK-backed server remains the more complete implementation

[chatgpt-codex-connector]

Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits.
Credits must be used to enable repository wide code reviews.

[mizoz]

Tested locally against Wave Terminal on Linux with an existing Codex OAuth login.

What worked:

• npm run wave:server exposed a local OpenAI-compatible endpoint on localhost:8080
• GET /health returned successfully
• POST /v1/chat/completions returned a valid OpenAI-style response
• SSE streaming returned data: ... chunks followed by [DONE]
• Wave picked up the custom mode from ~/.config/waveterm/waveai.json and was able to use the bridge without changing the existing default mode

Why this path exists:

• Wave already supports custom OpenAI-compatible chat endpoints
• there is no native Codex OAuth mode in Wave
• the thinner CLI-backed bridge is a practical local integration path, and it also helps when the SDK-backed server does not start cleanly in a given environment

This is intentionally minimal and local-first. The existing SDK-backed bridge remains the more complete implementation.