docs(tutorials): Pi 5 + Hailo cluster rvcsi tutorial

[ruvnet]

Summary

Adds docs/tutorials/pi5-cluster-cognitive-rf-observer.md — a
field-tested walkthrough for building a 4-node Raspberry Pi 5 + 2×
Hailo-8 multistatic Wi-Fi CSI cognitive RF observer using rvcsi.

Built against the v0-appliance v0.5.0-cognitive-rf-observer milestone:

• 4× Pi 5 + 2× Hailo-8 → CSI → 128-d embeddings on-device → cosine
  k-means w/ warm-start → 2nd-order Markov chain → SQLite brain corpus
• 446k+ fingerprints observed in development
• 16 stable RF states (IDs preserved across hourly re-clustering)
• 39% top-1 ceiling on 2nd-order next-state prediction (1.06× gain
  over 1st-order, 16× chance baseline)
• 34 cog-query subcommands, 4 JSON modes
• 12-category brain corpus, 15 expected services across 4 hosts

What's in the tutorial

1. Prerequisites + BOM (~$580 parts + workstation)
2. Architecture overview (ASCII topology diagram)
3. Per-node firmware: native ARM nexmon_csi build (cross-compile is a
   documented dead end)
4. Per-node services (per-host topology table)
5. Workstation pipeline (3 daemons, 7 timers, brain HTTP service)
6. Calibration: walk → cluster → warm-start IDs → Markov skill
7. cog-query CLI: 34 subcommands cheat-sheet
8. What you can measure
9. Troubleshooting table (painful lessons captured concretely)
10. Next steps

Companion gist

Detailed cookbook with every unit file, command, config:
https://gist.github.com/ruvnet/88e7b053c41cb4f4af7a7ec4af873017

Style fit

Mirrors existing docs/tutorials/cognitum-seed-pretraining.md format.

Test plan

• Render check on GitHub (tables, ASCII diagram alignment)
• Cross-links resolve (gist, v0-appliance ADR-206, nexmon_csi, hailo.ai)
• Spot-check recipe steps on a fresh Pi 5

🤖 Generated with claude-flow