LLM failures silently return empty results without user feedback

[sm-28601]

LLM failures are silently ignored in scan results

📌 Description

When LLM-based analysis (summary, impact, privacy) fails, the system catches exceptions and returns None values without informing the user.

The scan appears successful, but important sections are empty.

---

📷 Proof

📄 Code Reference

Example:

• File: workflow/nodes.py
• Exceptions are caught and values set to None

---

❗ Problem

• Users see empty sections with no explanation
• Misleading: scan appears successful
• Poor user experience

---

✅ Expected Behavior

• Show warning like:
  "Summary unavailable — LLM service temporarily failed"

---

💡 Suggested Fix

Add a warnings or partial_failures field in API response and display it in frontend.