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

navigator/mock_lbm_daemon.py

@@ -0,0 +1,47 @@
+# mock_lbm_daemon.py
+# Simple mock LBM daemon for Open Feed testing
+
+import zmq
+import json
+import time
+import math
+
+def mock_daemon():
+    ctx = zmq.Context()
+    pub = ctx.socket(zmq.PUB)
+    pub.bind("tcp://*:5556")
+    
+    print("[Mock LBM] Starting on port 5556...")
+    print("[Mock LBM] Simulating 1024x1024 grid with Khra'gixx signature")
+    
+    cycle = 0
+    while True:
+        # Simulate Khra'gixx wave: 64-cell + 16-cell harmonics
+        khra = math.sin(cycle * 0.02) * math.cos(cycle * 0.015) * 2.0
+        gixx = math.sin(cycle * 0.2) * 0.5
+        
+        coherence = 15.0 + khra + gixx
+        h64 = 7.8 + khra * 0.5
+        h32 = 0.01 + abs(gixx) * 0.1
+        vorticity = 0.5 + abs(khra) * 0.3
+        
+        data = {
+            "cycle": cycle,
+            "coherence": coherence,
+            "h64": h64,
+            "h32": h32,
+            "vorticity": vorticity,
+            "power_w": 50.0 + abs(khra) * 5.0,
+            "grid": 1024
+        }
+        
+        pub.send_json(data)
+        
+        if cycle % 100 == 0:
+            print(f"[Mock LBM] Cycle {cycle}: Coh={coherence:.3f}, H64={h64:.3f}")
+        
+        cycle += 1
+        time.sleep(0.01)  # 100Hz
+
+if __name__ == "__main__":
+    mock_daemon()