>  because running a zap-capable LN nodes is much more involved than running just an LN node.

Why is this still a case 2 years later?