add open-vocab YOLO-World engine with CLIP text encoder

- Add OpenVocab engine: YOLO-World ONNX detection with runtime
  vocabulary via CLIP text embeddings (two-input model support)
- Add header-only C++ CLIP BPE tokenizer (no external deps)
- Add detect(cv::Mat) overload to OpenVocab for camera-free testing
- Add standalone test binary (test_open_vocab) using test/assets images
- Add export scripts for YOLO-World and CLIP text encoder (uv/PEP 723)
- Add installation scripts for Linux x64, ARM (Jetson), and Windows
- Update CMakeLists to use find_package(onnxruntime) instead of FetchContent
- Update README and AGENTS with env vars, setup workflow, and test usage

scripts/install_linux_arm.sh

@@ -0,0 +1,75 @@
+#!/usr/bin/env bash
+# Install build dependencies for dora-zed on NVIDIA Jetson (ARM64 / L4T).
+# Skips any component that is already installed.
+# Uses aria2c for fast parallel downloads if available, otherwise falls back to wget.
+#
+# Supported JetPack versions for ZED SDK 5.x:
+#   JetPack 6.0 GA  (L4T 36.3) — Jetson Orin
+#   JetPack 6.1/6.2 (L4T 36.4)
+#   JetPack 6.2.2   (L4T 36.5)
+#   JetPack 7.0     (L4T 38.2) — Jetson Thor
+#   JetPack 7.1     (L4T 38.4)
+#
+# ONNX Runtime: Microsoft does not publish official GPU binaries for ARM64.
+#   The native engine (ZED SDK built-in TensorRT) works without ONNX Runtime.
+#   To use the open_vocab engine, build ONNX Runtime from source:
+#   https://onnxruntime.ai/docs/build/eps.html#tensorrt
+set -euo pipefail
+
+ZED_VERSION="5.2"
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+TMP_DIR="${SCRIPT_DIR}/../tmp"
+mkdir -p "$TMP_DIR"
+
+download() {
+    local url="$1" dest="$2"
+    if command -v aria2c &>/dev/null; then
+        aria2c -x 16 -s 16 -d "$(dirname "$dest")" -o "$(basename "$dest")" "$url"
+    else
+        wget -O "$dest" "$url"
+    fi
+}
+
+# --- Detect L4T version ---
+if [ ! -f /etc/nv_tegra_release ]; then
+    echo "Error: /etc/nv_tegra_release not found. Is this a Jetson device?" >&2
+    exit 1
+fi
+L4T_VERSION=$(head -1 /etc/nv_tegra_release | grep -oP 'R\K[0-9]+\.[0-9]+' | head -1)
+L4T_MAJOR=$(echo "$L4T_VERSION" | cut -d'.' -f1)
+echo ">>> Detected L4T ${L4T_VERSION} (major: ${L4T_MAJOR})"
+
+case "$L4T_MAJOR" in
+    36) L4T_PATH="l4t36" ;;
+    38) L4T_PATH="l4t38" ;;
+    35) L4T_PATH="l4t35" ;;
+    *)
+        echo "Warning: L4T ${L4T_MAJOR} not tested. Attempting l4t${L4T_MAJOR}." >&2
+        L4T_PATH="l4t${L4T_MAJOR}"
+        ;;
+esac
+
+# --- ZED SDK ---
+if [ -d /usr/local/zed ]; then
+    echo ">>> ZED SDK already installed at /usr/local/zed — skipping."
+else
+    echo ">>> Downloading ZED SDK ${ZED_VERSION} for Jetson (${L4T_PATH})..."
+    download \
+        "https://download.stereolabs.com/zedsdk/${ZED_VERSION}/${L4T_PATH}/jetsons" \
+        "${TMP_DIR}/ZED_SDK.run"
+    chmod +x "${TMP_DIR}/ZED_SDK.run"
+    echo ">>> Installing ZED SDK (silent)..."
+    sudo "${TMP_DIR}/ZED_SDK.run" -- silent skip_python skip_tools
+fi
+
+# --- ONNX Runtime ---
+echo ""
+echo ">>> ONNX Runtime (ARM64 GPU): no official pre-built binary available."
+echo "    The native ZED engine works without ONNX Runtime."
+echo "    To use the open_vocab engine, build from source:"
+echo "      https://onnxruntime.ai/docs/build/eps.html#tensorrt"
+echo ""
+
+echo ">>> Done. You can now run: mkdir -p build && cd build && cmake .. && make install"
+echo ">>> Downloaded files are in ${TMP_DIR}"