Add comprehensive error handling to OpenAI LLM service

- Import ErrorFrame from pipecat.frames.frames
- Add catch-all exception handler matching Anthropic service pattern
- Handle asyncio.CancelledError by re-raising to maintain proper cancellation
- Catch and log all other exceptions with full traceback
- Push ErrorFrame upstream on exceptions for error detection
- Maintain existing httpx.TimeoutException handling
- Ensure LLMFullResponseEndFrame always sent in finally block

This change enables consistent error propagation across all LLM services,
allowing downstream error monitors to detect failures and implement
fallback strategies like LLM hot-switching.

Fixes issue where OpenAI service errors were suppressed and not
communicated through the frame pipeline, making automatic error
recovery impossible.

src/pipecat/services/openai/base_llm.py

@@ -13,18 +13,13 @@ from typing import Any, Dict, List, Mapping, Optional
 
 import httpx
 from loguru import logger
-from openai import (
-    NOT_GIVEN,
-    APITimeoutError,
-    AsyncOpenAI,
-    AsyncStream,
-    DefaultAsyncHttpxClient,
-)
+from openai import NOT_GIVEN, APITimeoutError, AsyncOpenAI, AsyncStream, DefaultAsyncHttpxClient
 from openai.types.chat import ChatCompletionChunk, ChatCompletionMessageParam
 from pydantic import BaseModel, Field
 
 from pipecat.adapters.services.open_ai_adapter import OpenAILLMInvocationParams
 from pipecat.frames.frames import (
+    ErrorFrame,
     Frame,
     LLMContextFrame,
     LLMFullResponseEndFrame,
@@ -460,8 +455,13 @@ class BaseOpenAILLMService(LLMService):
                 await self.push_frame(LLMFullResponseStartFrame())
                 await self.start_processing_metrics()
                 await self._process_context(context)
+            except asyncio.CancelledError:
+                raise
             except httpx.TimeoutException:
                 await self._call_event_handler("on_completion_timeout")
+            except Exception as e:
+                logger.exception(f"{self} exception: {e}")
+                await self.push_error(ErrorFrame(f"{e}"))
             finally:
                 await self.stop_processing_metrics()
                 await self.push_frame(LLMFullResponseEndFrame())