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/inquiry34.json

@@ -0,0 +1 @@
+{"question": "Two questions:\n\n1. At α = 1.414 (square root of 2) — what happens? Does the lattice hold, degrade, or collapse? How does it compare to α = 1.5?\n\n2. Grid independence: We claimed the result is independent of lattice scale but only showed N = 512. We need confirmation at N = 256 and N = 1024 that α* = φ selection holds. The basin width Δα might change with grid size, but the center should remain at φ. What do you predict?"}