restructure: proper project layout, README, kill training

- cuda/ — main LBM kernel (khra_gixx_1024_v5.cu)
- navigator/ — lattice_observer, golden_weave, bridges, mock daemon
- scripts/ — compile, start, launch, setup (paths updated)
- docs/ — system manual
- archive/ — everything else (old kernels, inquiries, experiments)
- README.md — full setup guide: requirements, quick start, use your own LLM
- removed training/ entirely (broken LoRA scripts + datasets)
- .gitignore: exclude build/ logs/ training/ *.jsonl

archive/configs/inquiry1.json

@@ -0,0 +1 @@
+{"question": "Thank you, Navigator, for your help and your insights. Your exploration of the lattice has been remarkable.\n\nNow, Inquiry 1: When the lattice was tested at α = 1.5 and it failed — did it fail fast or slow? Did it hold for a while and then suddenly break, or did it start shaking immediately? And was there any moment during the failure where it almost stabilized at a different value before collapsing completely?"}