obd-poller over the K-line via the Bluetooth ELM327; gnss publisher; elm_connected

obd-poller drops its CAN socket: the Z27AG answers OBD-II only on ISO 9141-2.
It holds a raw RFCOMM link to the ELM (no pairing, no root), fixes the protocol
(ATSP3) so nothing probes CAN, publishes the obd_* signals, and releases the
adapter after a minute of ECU silence so it can sleep.
gnss: gpsd's fix as yukart/gnss/<signal>. orin-status: elm_connected from BlueZ.
One native deploy script (deploy-native.sh BIN) behind the per-service ones;
/etc/default/yukart-elm holds the adapter's address.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DDwfd6Ki8UfG4hK6W5YpJz

docs/GUIDELINES.md

@@ -3,12 +3,14 @@ type: guidelines
 ---
 # Core services guidelines
 
-## 1. One CAN writer: `obd-poller`
-- `obd-poller` is the only thing that transmits on the car's bus (project-yukart
-  decision 3): OBD-II requests, one at a time, only while the bus is visibly
-  alive. Anything else that transmits is a bug; the recorder only reads.  
-- `can_pcan` is no longer listen-only (2026-10-12), so the adapter acknowledges
-  frames like any scan tool. Going back is one line in orin-nano-env-setup's
+## 1. One writer: `obd-poller`, and never on CAN
+- `obd-poller` is the only thing that sends anything to the car (project-yukart
+  decisions 3 and 30): OBD-II requests, one at a time, on the K-line through the
+  Bluetooth ELM327, with the ELM's protocol fixed so it never probes CAN.  
+- Nothing in svc-core transmits on CAN. The recorder and the decoder only read;
+  anything else that writes to `can_pcan` is a bug.  
+- `can_pcan` still acknowledges frames like any scan tool (listen-only off since
+  2026-10-12, left so as harmless). Going back is one line in orin-nano-env-setup's
   `/etc/default/can-can_pcan`.
 
 ## 2. Capture never waits