feat: add chat_tui example for FastGPT with Textual interface

- Introduced a new example script `chat_tui.py` that provides a full-screen Textual interface for interacting with FastGPT.
- Implemented streaming chat updates, workflow logging, and modal handling for interactive nodes.
- Enhanced FastGPT client with new streaming capabilities and structured event types for better interaction handling.
- Normalized base URL handling in the client to prevent duplicate `/api` paths.
- Added tests for streaming event parsing and interaction handling.

fastgpt_client/async_client.py

@@ -44,9 +44,10 @@ class AsyncFastGPTClient(BaseClientMixin):
         # Initialize base client functionality
         super().__init__(api_key, base_url, timeout, max_retries, retry_delay, enable_logging)
 
+        connect_timeout = min(float(timeout), 15.0) if timeout and timeout > 0 else 15.0
         self._client = httpx.AsyncClient(
-            base_url=base_url,
-            timeout=httpx.Timeout(timeout, connect=5.0),
+            base_url=self.base_url,
+            timeout=httpx.Timeout(timeout, connect=connect_timeout),
         )
 
     async def __aenter__(self):
@@ -270,6 +271,9 @@ class AsyncFastGPTClient(BaseClientMixin):
         try:
             error_data = response.json()
             message = error_data.get("message", f"HTTP {response.status_code}")
+        except httpx.ResponseNotRead:
+            message = f"HTTP {response.status_code}"
+            error_data = None
         except (ValueError, KeyError, AttributeError):
             # If we can't parse JSON (e.g., streaming response or invalid JSON), use status code
             message = f"HTTP {response.status_code}"