feat(advisor): add 0.85 advisory threshold triggering LLM suggestions

- Add advisory_threshold=0.85 field to MetricRule (higher-is-better metrics)
- diagnose() now emits severity='low' for scores in (warning_threshold, 0.85)
- noise_sensitivity (lower-is-better) keeps its existing two-tier thresholds
- writer.py: severity labels mapped to Chinese (严重/警告/待优化)
- llm_analyzer.py: prompt explains low/warning/critical tiers in Chinese
- Tests: 5 new cases for 'low' severity, updated log summary assertions

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

rag_eval/advisor/rules.py

@@ -14,6 +14,9 @@ class MetricRule:
     higher_is_better: bool  # False for noise_sensitivity
     root_causes: list[str]
     suggested_actions: list[str]
+    # Scores below this threshold trigger a "low" advisory (LLM suggestion requested).
+    # Only applies to higher_is_better metrics; noise_sensitivity uses existing thresholds.
+    advisory_threshold: float = 0.85
 
 
 METRIC_RULES: dict[str, MetricRule] = {
@@ -208,10 +211,14 @@ def diagnose(
             elif mean < rule.warning_threshold:
                 severity = "warning"
                 threshold = rule.warning_threshold
+            elif mean < rule.advisory_threshold:
+                # Score is acceptable but below 0.85 — request LLM optimization advice.
+                severity = "low"
+                threshold = rule.advisory_threshold
             else:
-                continue  # above warning threshold → no diagnosis
+                continue  # >= advisory_threshold → no diagnosis needed
         else:
-            # lower is better (noise_sensitivity)
+            # lower is better (noise_sensitivity): keep existing two-tier logic
             if mean > rule.critical_threshold:
                 severity = "critical"
                 threshold = rule.critical_threshold