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/misc/send_cmd.py

@@ -0,0 +1,30 @@
+import zmq, time, json, sys
+ctx = zmq.Context()
+
+# Also subscribe to ACK channel (5559) to confirm command receipt
+ack = ctx.socket(zmq.SUB)
+ack.setsockopt_string(zmq.SUBSCRIBE, "")
+ack.connect("tcp://127.0.0.1:5559")
+
+s = ctx.socket(zmq.PUB)
+s.connect("tcp://127.0.0.1:5557")
+time.sleep(0.5)
+
+cmd = json.dumps({"cmd": "load_state", "path": "/mnt/d/fractal-brain/beast-build/ckpt_20260319_133430_c7521391.bin"})
+s.send_string(cmd)
+print(f"Sent: {cmd}")
+
+# Wait for ACK
+poller = zmq.Poller()
+poller.register(ack, zmq.POLLIN)
+events = dict(poller.poll(5000))
+if ack in events:
+    reply = ack.recv_string()
+    print(f"ACK: {reply}")
+else:
+    print("No ACK received within 5s (check v4 logs)")
+
+s.close()
+ack.close()
+ctx.term()
+print("Done")