fix: resolve dashboard websocket 'disconnected' status

- Fixed status indicator UI mapping in websocket.js and index.html.
- Added missing CSS for connection status indicator and pulse animation.
- Made initial model registry fetch asynchronous to prevent blocking server startup.
- Improved configuration loading to correctly handle LLM_PROXY__SERVER__PORT from environment.

internal/server/server.go

@@ -32,13 +32,6 @@ type Server struct {
 func NewServer(cfg *config.Config, database *db.DB) *Server {
 	router := gin.Default()
 	hub := NewHub()
-	
-	// Fetch registry (non-blocking for startup if it fails, but we'll try once)
-	registry, err := utils.FetchRegistry()
-	if err != nil {
-		fmt.Printf("Warning: Failed to fetch initial model registry: %v\n", err)
-		registry = &models.ModelRegistry{Providers: make(map[string]models.ProviderInfo)}
-	}
 
 	s := &Server{
 		router:    router,
@@ -48,9 +41,19 @@ func NewServer(cfg *config.Config, database *db.DB) *Server {
 		sessions:  NewSessionManager(cfg.KeyBytes, 24*time.Hour),
 		hub:       hub,
 		logger:    NewRequestLogger(database, hub),
-		registry:  registry,
+		registry:  &models.ModelRegistry{Providers: make(map[string]models.ProviderInfo)},
 	}
 
+	// Fetch registry in background
+	go func() {
+		registry, err := utils.FetchRegistry()
+		if err != nil {
+			fmt.Printf("Warning: Failed to fetch initial model registry: %v\n", err)
+		} else {
+			s.registry = registry
+		}
+	}()
+
 	// Initialize providers
 	if cfg.Providers.OpenAI.Enabled {
 		apiKey, _ := cfg.GetAPIKey("openai")