audit(tracking): add wolf-pack runtime detection to fleet health probe

Issue #500 cross-audit discovered six untracked wolf-* processes running
under /tmp/wolf-pack/ that were not reflected in systemd or fleet health
dashboards. This change adds detection to the automated health probe.

Change:
  scripts/fleet_health_probe.sh — new 'Untracked Wolf-Pack Runtimes'
  section that pgrep's for 'wolf-[0-9]' patterns and logs a WARNING
  with the count when found. The check is informational only and does
  not fail the health probe (status remains 0).

Smoke test:
  bash -n scripts/fleet_health_probe.sh  # syntax OK
  Script runs successfully with writable LOG_DIR/HEARTBEAT_DIR overrides.

This is the smallest concrete fix implementing the tracking part of
issue #500's action item 4 (Audit and track wolf pack runtime).

Closes #500

scripts/fleet_health_probe.sh

@@ -71,6 +71,15 @@ for proc in $CRITICAL_PROCESSES; do
     fi
 done
 
+# --- Untracked Wolf-Pack Runtimes ---
+# Detect any wolf-* processes that are not managed by systemd/fleet tracking.
+# These processes exist under /tmp/wolf-pack/ and should appear in health logs.
+if pgrep -f "wolf-[0-9]" >/dev/null 2>&1; then
+    wolf_count=$(pgrep -f "wolf-[0-9]" | wc -l | tr -d ' ')
+    log "WARNING: Untracked wolf-pack runtime detected — ${wolf_count} active processes (not in systemd/fleet tracking)"
+    # Not marked as failure — informational only for now
+fi
+
 # --- Heartbeat Touch ---
 touch "${HEARTBEAT_DIR}/fleet_health.last"