Rebrand CommGate to gsmnode and adopt new design system

Rename the whole project from CommGate to gsmnode and re-skin every
surface onto the new signal-green + ink design system (Space Grotesk /
IBM Plex Sans / JetBrains Mono, flat surfaces, two-arrow routing mark,
lowercase gsm+node wordmark).

- Web App + API panel: rewrite token layer, gn-* utilities, data-gsm-theme,
  new logo/favicon assets; both builds verified.
- Phone App (Flutter): green theme, new mark/wordmark widgets, adaptive
  launcher icons; rename Android applicationId/package to app.gsmnode.phone;
  bump AGP to 8.6.0 and google_fonts to 8.1.0; debug APK build verified.
- API Server (Go): panel routes, User-Agent, health service id, branding.
- Home Assistant Plugin: rename integration domain sms_gateway to gsmnode
  (folder, DOMAIN, services, entity ids, classes); py_compile verified.
- Update all READMEs; add .gitignore (excludes Design/, build artifacts).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

API Server/internal/api/sweeper.go

@@ -0,0 +1,74 @@
+package api
+
+import (
+	"context"
+	"log"
+	"time"
+
+	"smsgateway/apiserver/internal/pb"
+)
+
+// sweepInterval is how often stale messages are checked.
+const sweepInterval = 30 * time.Second
+
+// StartExpiryWorker launches a background loop that fails messages/calls which
+// no device picked up or reported on within the configured MessageTTL. Without
+// it, a message targeting an offline device would stay Pending forever.
+func (s *Server) StartExpiryWorker(ctx context.Context) {
+	go func() {
+		ticker := time.NewTicker(sweepInterval)
+		defer ticker.Stop()
+		// Run once at startup so stale items clear promptly.
+		s.expireStaleMessages(ctx)
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			case <-ticker.C:
+				s.expireStaleMessages(ctx)
+			}
+		}
+	}()
+}
+
+// expireStaleMessages marks Pending/Processed messages older than MessageTTL as
+// Failed. "Pending" means no online device pulled it; "Processed" means a device
+// pulled it but never reported a terminal state (e.g. it crashed).
+func (s *Server) expireStaleMessages(ctx context.Context) {
+	cutoff := time.Now().UTC().Add(-s.cfg.MessageTTL).Format("2006-01-02 15:04:05.000Z")
+	filter := `(status = "` + statusPending + `" || status = "` + statusProcessed +
+		`") && updated < "` + cutoff + `"`
+
+	res, err := s.pb.List(ctx, colMessages, pb.ListOptions{Filter: filter, PerPage: 200})
+	if err != nil {
+		log.Printf("expiry sweep failed: %v", err)
+		return
+	}
+
+	expired := 0
+	for _, rec := range res.Items {
+		// Don't expire messages scheduled for the future.
+		if asString(rec["schedule_at"]) != "" {
+			continue
+		}
+		id := asString(rec["id"])
+		const reason = "expired: no device processed the message within the timeout"
+		if _, err := s.pb.Update(ctx, colMessages, id, pb.Record{
+			"status": statusFailed,
+			"error":  reason,
+		}); err != nil {
+			log.Printf("expire message %s: %v", id, err)
+			continue
+		}
+		expired++
+		s.dispatchWebhooks(asString(rec["owner"]), asString(rec["device"]), "sms:failed", map[string]any{
+			"message_id": id,
+			"type":       asString(rec["type"]),
+			"status":     statusFailed,
+			"error":      reason,
+		})
+	}
+	if expired > 0 {
+		log.Printf("expired %d stale message(s) older than %s", expired, s.cfg.MessageTTL)
+	}
+}