feat: RDG session flap rule, qwinsta/logoff queue and dashboard UI (0.9.13)

Detect 302→303 within 1–10s, Problem with 30s dedup, rdg_flap flag.
Agent command poll API, admin qwinsta/logoff from Overview. Migration 016.

backend/app/services/problems.py

@@ -1,6 +1,6 @@
 """Auto-create Problems from ingested events (rules + correlation)."""
 
-from datetime import datetime, timezone
+from datetime import datetime, timedelta, timezone
 
 from sqlalchemy import select
 from sqlalchemy.orm import Session
@@ -11,6 +11,7 @@ from app.services.host_health import HEARTBEAT_TYPE
 from app.services.problem_rules import (
     RuleMatch,
     RULE_HOST_SILENCE,
+    RULE_RDG_SESSION_FLAP,
     build_fingerprint,
     pick_rule_match,
     resolve_host_silence_problems,
@@ -18,8 +19,6 @@ from app.services.problem_rules import (
 
 
 def _correlation_cutoff(now: datetime) -> datetime:
-    from datetime import timedelta
-
     minutes = get_settings().sac_problem_correlation_window_minutes
     return now - timedelta(minutes=minutes)
 
@@ -66,7 +65,10 @@ def open_or_append_problem(
         match.fingerprint_suffix,
     )
     now = datetime.now(timezone.utc)
-    cutoff = _correlation_cutoff(now)
+    if match.rule_id == RULE_RDG_SESSION_FLAP:
+        cutoff = now - timedelta(seconds=get_settings().sac_rdg_flap_dedup_sec)
+    else:
+        cutoff = _correlation_cutoff(now)
 
     open_problem = db.scalar(
         select(Problem)