fix: Fix SDK async streaming - use create(stream=True) instead of stream()

The Anthropic SDK's messages.stream() is a synchronous context manager,
not async. For async streaming, we need to use messages.create(..., stream=True)
which returns an async iterator of ServerSentEvent objects.

Changed from:
    async with client.messages.stream(**request_kwargs) as stream:
To:
    stream = await client.messages.create(**request_kwargs, stream=True)
    async for event in stream:

aisbf/providers.py

@@ -3932,10 +3932,10 @@ class ClaudeProviderHandler(BaseProviderHandler):
     
     async def _handle_streaming_request_sdk(self, client, request_kwargs: Dict, model: str):
         """
-        Handle streaming request using Anthropic SDK.
+        Handle streaming request using Anthropic SDK's async streaming API.
         
-        The SDK handles proper streaming event parsing and retries.
-        We convert SDK events to OpenAI-compatible SSE chunks.
+        Uses client.messages.create(..., stream=True) which returns an async iterator
+        of ServerSentEvent objects. We parse these events and convert to OpenAI SSE chunks.
         """
         import logging
         import json
@@ -3961,8 +3961,9 @@ class ClaudeProviderHandler(BaseProviderHandler):
         idle_timeout = self.stream_idle_timeout
         
         try:
-            # Use SDK's streaming API with context manager
-            async with client.messages.stream(**request_kwargs) as stream:
+            # Use SDK's async streaming API - create(stream=True) returns async iterator
+            stream = await client.messages.create(**request_kwargs, stream=True)
+            
             async for event in stream:
                 # Update idle watchdog
                 last_event_time = time.time()