feat: add thinking_config to AnthropicClient whitelist

Updated the allowed_keys tuple to permit thinking_config parameters. This ensures that the extended reasoning budget defined in the trading graph is actually passed to the Anthropic API.

tradingagents/llm_clients/anthropic_client.py

@@ -16,7 +16,7 @@ class AnthropicClient(BaseLLMClient):
         """Return configured ChatAnthropic instance."""
         llm_kwargs = {"model": self.model}
 
-        for key in ("timeout", "max_retries", "api_key", "max_tokens", "callbacks", "http_client", "http_async_client"):
+        for key in ("timeout", "max_retries", "api_key", "max_tokens", "callbacks", "http_client", "http_async_client", "thinking_config"):
             if key in self.kwargs:
                 llm_kwargs[key] = self.kwargs[key]