Adapter for the ChatGPT Codex backend exposed by ReqLLM.
The provider uses LLMProxy's token pool for OAuth tokens and delegates Codex
request construction, OAuth account handling, and WebSocket streaming to
ReqLLM.Providers.OpenAICodex. It converts ReqLLM's canonical response and
stream chunk structs back into the OpenAI-compatible wire shapes that LLMProxy
HTTP routes expose.