refactor: parallel LLM analysis via asyncio.gather instead of sequential

guarddog_nexus/core/harvester.py

@@ -228,12 +228,15 @@ async def _run_llm_analysis(findings: list[Finding], session: AsyncSession) -> l
     await session.commit()
 
     reports = []
-    for finding in findings:
+
+    async def _analyze_one(finding):
         report = await analyze_finding(finding.data)
         if report:
             finding.report = report
             reports.append(report)
         else:
             finding.report = None
+
+    await asyncio.gather(*(_analyze_one(f) for f in findings))
     await session.commit()
     return reports