Fix field name mismatches, add events filter, in-place suppression refresh

- links.html: fix all field name bugs (auto_negotiation→autoneg, full_duplex,
  tx/rx_errors/drops_per_sec→_rate, tx/rx_bytes_per_sec→_rate, poe_total_w/poe_max_w
  computed from ports, renderUnifiSwitches uses top-level updated timestamp)
- suppressions.html: in-place DOM refresh after create/remove (no page reload),
  datalist autocomplete for target names, form reset after submit
- inspector.html: ESC key closes detail panel via lt.keys.on
- index.html: events filter bar with search input + severity pills (All/Critical/Warning),
  MutationObserver re-applies filter after dynamic updates
- style.css: g-section-actions, events-filter-bar, sev-pills layout
- app.js/db.py/monitor.py: carry forward prior session fixes (Promise.allSettled,
  daemon_ok, stale connection handling, double Prometheus call, self.cfg fix)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

monitor.py

@@ -325,6 +325,7 @@ class LinkStatsCollector:
 
     def __init__(self, cfg: dict, prom: 'PrometheusClient',
                  unifi: Optional['UnifiClient'] = None):
+        self.cfg   = cfg
         self.prom  = prom
         self.pulse = PulseClient(cfg)
         self.unifi = unifi
@@ -876,8 +877,7 @@ class NetworkMonitor:
     # ------------------------------------------------------------------
     # Snapshot collection (for dashboard)
     # ------------------------------------------------------------------
-    def _collect_snapshot(self) -> dict:
-        iface_states = self.prom.get_interface_states()
+    def _collect_snapshot(self, iface_states: Dict[str, Dict[str, bool]]) -> dict:
         unifi_devices = self.unifi.get_devices() or []
 
         hosts = {}
@@ -930,23 +930,23 @@ class NetworkMonitor:
             try:
                 logger.info('Starting network check cycle')
 
-                # 1. Collect and store snapshot for dashboard
-                snapshot = self._collect_snapshot()
+                # 1. Fetch interface states once — shared by snapshot and alert processing
+                iface_states = self.prom.get_interface_states()
+
+                # 2. Collect and store snapshot for dashboard
+                snapshot = self._collect_snapshot(iface_states)
                 db.set_state('network_snapshot', snapshot)
                 db.set_state('last_check', _now_utc())
 
-                # 2. Collect link stats (ethtool + traffic metrics)
+                # 3. Collect link stats (ethtool + traffic metrics)
                 try:
                     link_data = self.link_stats.collect(self._instance_map)
                     db.set_state('link_stats', link_data)
                 except Exception as e:
                     logger.error(f'Link stats collection failed: {e}', exc_info=True)
 
-                # 3. Process alerts (separate Prometheus call for fresh data)
-                # Load suppressions once per cycle to avoid N*M DB queries
+                # 4. Process alerts using already-fetched interface states
                 suppressions = db.get_active_suppressions()
-
-                iface_states = self.prom.get_interface_states()
                 self._process_interfaces(iface_states, suppressions)
 
                 unifi_devices = self.unifi.get_devices()