docs: README — data is wired to live MariaDB + reject writes (was "currently SAMPLE")

Overview reads live MariaDB and POST /reject writes since 2026-07-02. Replaced the
"wire it (TODO)" section with the live wiring; noted the HTTP-200+{success:false}
contract and the rejTtId column added by fastpress migration 009; collapsed the
resolved open questions (per-ttId booking + belt semantics), leaving only the
login-free-write security question open.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

README.md

@@ -90,32 +90,33 @@ Belt colour mapping (matches PressV process colours): `notYetSent`→violet,
 `wagen`→cyan, `unterwegs`→amber, `ofen`→orange, `fertig`→emerald; the reject accent
 is Grafana dark-red `#C4162A`.
 
-### Wiring the REAL data (currently SAMPLE)
+### Live data wiring
 
-`api/routes.py` returns sample data and `POST /reject` echoes success **without
+`api/routes.py` reads live MariaDB and `POST /reject` writes (details in the gateway
-writing**. To make it live (see the `TODO` there + `.claude/memory/ausschuss-station.md`):
+memory `.claude/memory/ausschuss-station.md`; `sample:true` only in the standalone dev
+fallback). Via `from .._db import run_select_query, transaction`:
 
-- `from .._db import run_select_query, transaction`.
+- **overview:** reads `production.temper_tracking` across **all** presses (un-scoped),
-- **overview:** read `production.temper_tracking` across **all** presses (un-scoped),
+  one row per `ttId`, newest-first by `ttOutOfOvenAt` (`ttStatus='done'`), joined to
-  one row per `ttId`, newest first by `ttOutOfOvenAt` (`ttStatus='done'`); join
+  `active_orders`/`finished_orders`, windowed on `ttOutOfOvenAt`. The order-level
-  `active_orders`/`finished_orders` for the article. Add each order's 5-station
+  5-station distribution reuses `modules/fastpress/press/temper.py::_temper_aggregates_all`.
-  distribution — reuse `modules/fastpress/press/temper.py::_temper_aggregates_all`
-  (notYetSent / sentToTemper / onWayToOven / inOven / finishedTempering). Filter by
-  `window` on `ttOutOfOvenAt`.
 - **reasons:** `SELECT id, reason_de, reason_en FROM settings.reject_reasons WHERE active=1`.
-- **reject:** one `INSERT` into `production.rejects` inside a `transaction()` — same
+- **reject:** one `INSERT` into `production.rejects` inside a `transaction()` (an
-  shape as `modules/fastpress/press/rejects.py::send_reject`. Booking references the
+  un-scoped reuse of PressV `send_reject`), keyed by `rejOrdernumber`+`rejPress` from
-  cart, which needs an **additive nullable `rejTtId` column** (a DBA applies it; the
+  the body — the device-scoped `POST /press/rejects/send` can't be reused (it derives
-  runtime account has no `ALTER`). The device-scoped `POST /press/rejects/send`
+  the press from the device JWT). ⚠ `/reject` returns **HTTP 200 + `{success:false}`**
-  can't be reused — it derives the press from the device JWT; here `press`+`ttId`
+  on failure (not 4xx/5xx — else the frontend store swallows it into a fake success).
-  come from the body.
 
-### Open questions before go-live (from the design memory)
+Note: a nullable `rejTtId` column was later added (fastpress migration 009) so rejects
+attribute to the correct press cycle — see the gateway memory
+`temper-cycle-reject-attribution.md`.
 
-1. **Security of the login-free write:** network/reverse-proxy restriction only, or a
+### Open question before go-live
-   static `STATION_TOKEN` in `.env`? (No TLS/reverse-proxy exists yet.)
+
-2. Confirm the `rejTtId` column name + that per-`ttId` booking is desired (yes, per spec).
+**Security of the login-free write:** network/reverse-proxy restriction only, or a
-3. Belt semantics: order-level distribution (chosen) vs. the cart's own journey position.
+static `STATION_TOKEN` in `.env`? (No TLS/reverse-proxy exists yet.) Because this
+terminal **writes**, resolve this before exposing it beyond a trusted LAN. (Resolved:
+per-`ttId` booking is desired; belt shows the order-level distribution.)
 
 ---