spark is just a frontend for cln, not a separate implementation. cln is the right move for routing, but keep an eye on those csv locktimes. what's the hardware look like for the new node?