Add max_tokens=1024 to LLM analysis calls

The analyze_message and raw_analyze methods had no max_tokens limit,
causing thinking models (Qwen3-VL-32B-Thinking) to generate unlimited
reasoning tokens before responding — taking 5+ minutes per message.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

utils/llm_client.py

@@ -123,6 +123,7 @@ class LLMClient:
                     tools=[ANALYSIS_TOOL],
                     tool_choice={"type": "function", "function": {"name": "report_analysis"}},
                     temperature=0.1,
+                    max_tokens=1024,
                 )
 
                 choice = response.choices[0]
@@ -255,6 +256,7 @@ class LLMClient:
                     tools=[ANALYSIS_TOOL],
                     tool_choice={"type": "function", "function": {"name": "report_analysis"}},
                     temperature=0.1,
+                    max_tokens=1024,
                 )
 
                 choice = response.choices[0]