refactor(reasoning)!: remove Node-RED engine and integrations

BREAKING CHANGE: NODE_RED reasoning is no longer supported. Unsupported
mode values fall back to PROXY. PROXY and OPENAI remain available.

s2s-client/s2s_client/models/requests.py

@@ -20,7 +20,7 @@ class ReasoningRequest(BaseModel):
     # Optional phone number. When S2S_INJECT_PROFILE is true the proxy fetches
     # the caller profile from VOCERO_USER_PHONE_AUTH_URL and:
     #   1. prepends the serialized profile to user_context before feeding the
-    #      engine (NODE_RED or OPENAI);
+    #      OpenAI-compatible provider;
     #   2. appends the profile context to the LLM instructions prompt.
     phoneNumber: str | None = None
     # Optional podcast episode ID. When set the proxy fetches the episode from

s2s-client/s2s_client/models/responses.py

@@ -21,7 +21,7 @@ class ConfigResponse(OpenResponse):
     allowDirect: bool
     s2sUrl: str
     primaryPrompt: str
-    reasoningType: Literal["PROXY", "OPENAI", "NODE_RED"]
+    reasoningType: Literal["PROXY", "OPENAI"]
     injectProfile: bool
     debugPrompt: bool
     startSessionMode: str
@@ -69,13 +69,7 @@ class ToolInvocation(OpenResponse):
 
 class ReasoningResponse(OpenResponse):
     answer: str
-    log_only: bool | None = None
-    usage: Any = None
     usage_pass2: Any = None
-    latency_ms: Any = None
-    intent: Any = None
-    caller_profile_keys: Any = None
-    debug: Any = None
     apipilot_invoked: ToolInvocation | None = None