pub fn decode_openai_stream_with_limits( runner: Symbol, model: &str, body: &[u8], include_raw: bool, limits: DecodeLimits, ) -> Result<Expr>
Decodes OpenAI chat-completion SSE under caller-supplied limits.