Switch primary model to flash to fix GCP LB 30s timeout

gemini-3.1-pro-preview takes ~25s per call, hitting the GCP load
balancer's 30s hard timeout before analysis completes. Flash model
returns in ~5-8s, fitting comfortably within the limit. Pro model
kept as fallback.

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

backend/app/services/gemini_service.py

@@ -38,8 +38,8 @@ class GeminiService:
             api_key=api_key,
             http_options={"timeout": _FALLBACK_TIMEOUT_MS},
         )
-        self.model = "gemini-3.1-pro-preview"
-        self.fallback_model = "gemini-3-flash-preview"
+        self.model = "gemini-3-flash-preview"
+        self.fallback_model = "gemini-3.1-pro-preview"
 
     async def _generate_content(
         self,