Bump to Qwen3-14B for better tool-output grounding

Qwen3-8B correctly invoked every tool and saw the real prices from
yfinance but then synthesized the final signal from training-data
memory, producing a $195 entry for AAPL instead of the tool-reported
$293. Qwen3-14B is materially better at grounding final numbers in
the tool outputs it just observed. The MI300X (192 GB) fits 14B with
plenty of headroom.

crew/config.py

@@ -8,7 +8,7 @@ class LLMConfig:
     """Configuration for the vLLM endpoint connection."""
 
     base_url: str = "http://localhost:8000/v1"
-    model_name: str = "Qwen/Qwen3-8B"
+    model_name: str = "Qwen/Qwen3-14B"
     temperature: float = 0.7
     max_tokens: int = 1024
     request_timeout: int = 120  # seconds