Add containerized OCR pipeline for fuel receipts

Self-hosted, verifiable pipeline for extracting structured data (station,
date/time, price per litre, litres, discount, total) from German fuel
receipt photos: grayscale + mild bilateral filter + contrast stretch,
Tesseract OCR, then LLM structuring via a small local model (Qwen 2.5,
served by Ollama). Each step is independently runnable and inspectable
through numbered scripts against a shared tooling container.

Ollama runs as its own long-lived container (0_run_ollama.sh), separate
from the stateless per-receipt pipeline steps, so the model stays loaded
in memory between runs. Includes a containerized ground-truth batch test
(test_batch.sh) that reproduces the full pipeline over the test receipts
and reports per-field accuracy.

Targets eventual deployment on a Raspberry Pi 4 k3s cluster.

0_run_ollama.sh

@@ -0,0 +1,64 @@
+#!/usr/bin/env bash
+# Step 0: start the Ollama inference server that the pipeline's
+# structuring step (3_extract_data.sh) talks to. Runs the official
+# ollama/ollama image as a long-lived container with model weights kept
+# in a named Docker volume - deliberately separate from the
+# tooling-container (which is stateless and one-shot per receipt), so the
+# model stays loaded in memory between pipeline runs instead of being
+# reloaded on every receipt.
+#
+# Usage: ./0_run_ollama.sh [-model <name>]
+set -euo pipefail
+
+cd "$(dirname "${BASH_SOURCE[0]}")"
+
+fail() { echo "FAIL: $*" >&2; exit 1; }
+
+command -v docker >/dev/null 2>&1 || fail "docker not found in PATH"
+
+MODEL="qwen2.5:3b"
+while [ $# -gt 0 ]; do
+    case "$1" in
+        -model) MODEL="${2:-}"; shift 2 ;;
+        *) fail "unknown argument: $1" ;;
+    esac
+done
+
+CONTAINER=gastronomy-ollama
+VOLUME=gastronomy-ollama-data
+PORT=11434
+
+if docker ps --format '{{.Names}}' | grep -qx "$CONTAINER"; then
+    echo "== $CONTAINER already running =="
+elif docker ps -a --format '{{.Names}}' | grep -qx "$CONTAINER"; then
+    echo "== Starting existing $CONTAINER container =="
+    docker start "$CONTAINER" >/dev/null
+else
+    echo "== Creating $CONTAINER (model weights kept in volume $VOLUME) =="
+    docker run -d \
+        --name "$CONTAINER" \
+        -p "127.0.0.1:$PORT:11434" \
+        -v "$VOLUME:/root/.ollama" \
+        --restart unless-stopped \
+        ollama/ollama >/dev/null
+fi
+
+echo "== Waiting for Ollama to become ready =="
+ready=0
+for _ in $(seq 1 30); do
+    if curl -s --max-time 2 "http://127.0.0.1:$PORT/api/tags" >/dev/null 2>&1; then
+        ready=1
+        break
+    fi
+    sleep 1
+done
+[ "$ready" = "1" ] || fail "Ollama did not become ready on port $PORT"
+
+if ! docker exec "$CONTAINER" ollama list | awk 'NR>1{print $1}' | grep -qx "$MODEL"; then
+    echo "== Pulling model $MODEL =="
+    docker exec "$CONTAINER" ollama pull "$MODEL"
+fi
+
+echo "== Done =="
+echo "Ollama running at http://127.0.0.1:$PORT (model: $MODEL)"
+echo "Next: ./1_preprocess_image.sh -in <photo> -out <scratch-dir>"