refactor: solidify startup UX and engine-aware preprocessing

- Slim core deps: move ML stack to optional extras (paddle/tables/figures/scientific/full)
- Lazy settings proxy with config search paths (env var, cwd, user dir)
- New commands: init, demo, setup [basic|full], first-run guard on run/watch
- Engine-aware preprocessing: Paddle gets original image (fixes dark mode 0.83->0.95)
- Results table shows Skipped count; lazy run-dir creation
- kg_ocr marked experimental with extra, Docker defaults with OCR_PIPELINE_CONFIG
- 25/25 tests, ruff clean

src/ocr_pipeline/onboarding.py

@@ -0,0 +1,64 @@
+from __future__ import annotations
+
+import platform
+import shutil
+from pathlib import Path
+
+import yaml
+
+from ocr_pipeline.config import DEFAULT_CONFIG_FILENAME, user_config_dir
+from ocr_pipeline.utils.logging import get_logger
+
+logger = get_logger(__name__)
+
+CONFIG_HEADER = "# ocr-pipeline configuration (generated by `ocr-pipeline init`)\n"
+
+
+def default_screenshots_dir() -> Path:
+    """The directory where this OS puts screenshots by default."""
+    system = platform.system().lower()
+    if system == "darwin":
+        return Path.home() / "Desktop"
+    if system == "windows":
+        return Path.home() / "Pictures" / "Screenshots"
+    return Path.home() / "Pictures"
+
+
+def tesseract_path() -> str | None:
+    """Path to the tesseract executable, or None when it is not on PATH."""
+    return shutil.which("tesseract")
+
+
+def tesseract_install_hint() -> str:
+    """One-line install instruction for the current platform."""
+    system = platform.system().lower()
+    if system == "darwin":
+        return "brew install tesseract"
+    if system == "windows":
+        return "choco install tesseract  # or https://github.com/UB-Mannheim/tesseract/wiki"
+    return "sudo apt-get install tesseract-ocr  # or your distro's package manager"
+
+
+def default_config(target_dir: Path, screenshots_dir: Path | None = None) -> dict:
+    """Minimal config content: safe Tesseract defaults, everything optional off."""
+    shots = screenshots_dir or default_screenshots_dir()
+    return {
+        "input": {"paths": [str(shots)]},
+        "ocr": {"engine": "tesseract"},
+        "output": {"write_consolidated": True},
+    }
+
+
+def write_default_config(target_dir: Path, screenshots_dir: Path | None = None) -> Path:
+    """Write a minimal config.yaml into target_dir and return its path."""
+    target_dir.mkdir(parents=True, exist_ok=True)
+    config_path = target_dir / DEFAULT_CONFIG_FILENAME
+    body = yaml.safe_dump(default_config(target_dir, screenshots_dir), sort_keys=False)
+    config_path.write_text(CONFIG_HEADER + body, encoding="utf-8")
+    logger.info("config_written", path=str(config_path))
+    return config_path
+
+
+def default_config_location(local: bool) -> Path:
+    """Where `init` writes config: cwd with --local, else the per-user config dir."""
+    return Path.cwd() if local else user_config_dir()