fix: unreliable websocket functionality (#144)

* increases timeout

* refactor: reuse constants for redis db

* avoid closing connection on pong error

* fix: prevent electron from pausing timer executions

Our pings could be paused when our app is backgrounded.

* refactor: remove unnecessary refs

* fix: presence unreliable within same pod

It was broadcasting presence changes to other pods, but not notifying
clients connected to the same pod

go/internal/pusher/server.go

@@ -15,14 +15,17 @@ import (
 type Server struct {
 	pbpusher.UnimplementedPusherServiceServer
 
+	ctx     context.Context // server-scoped context for graceful shutdown
 	hub     *Hub
 	bridge  *RedisBridge
 	authSvc auth.AuthService
 }
 
-// NewServer creates a new pusher server.
-func NewServer(hub *Hub, bridge *RedisBridge, authSvc auth.AuthService) *Server {
+// NewServer creates a new pusher server. The ctx controls the lifetime of all
+// WebSocket connections — when cancelled, all connections are closed gracefully.
+func NewServer(ctx context.Context, hub *Hub, bridge *RedisBridge, authSvc auth.AuthService) *Server {
 	return &Server{
+		ctx:     ctx,
 		hub:     hub,
 		bridge:  bridge,
 		authSvc: authSvc,
@@ -59,7 +62,10 @@ func (s *Server) HandleWebSocket(w http.ResponseWriter, r *http.Request) {
 
 	slog.Info("websocket connected", "connId", connID, "humanId", session.HumanId)
 
-	ctx, cancel := context.WithCancel(r.Context())
+	// Use server context, NOT r.Context(). After WebSocket upgrade, the HTTP
+	// request context can be cancelled by load balancers or Go's HTTP server,
+	// and nhooyr/websocket permanently closes the conn on any context error.
+	ctx, cancel := context.WithCancel(s.ctx)
 	defer cancel()
 
 	// Auto-subscribe to presence channel so this user appears online