improve the OCR pipeline processing and outputs formatting

tests/test_output.py

@@ -0,0 +1,29 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from ocr_pipeline.output.markdown import MarkdownWriter
+
+
+def test_writer_uses_metadata_and_unique_source_hash(tmp_path: Path, monkeypatch) -> None:
+    from ocr_pipeline.output import markdown
+
+    monkeypatch.setattr(markdown.settings.output, "base_directory", str(tmp_path))
+    monkeypatch.setattr(markdown.settings.output, "organize_by", "date_run")
+    writer = MarkdownWriter()
+    writer.start_run("2026-07-19T12:00:00+00:00", 2)
+    metadata = {
+        "source_path": "/screenshots/SCR-001.png",
+        "source_hash": "0123456789abcdef",
+        "timestamp": "2026-07-19T12:00:00+00:00",
+        "chunk_index": 0,
+        "total_chunks": 1,
+    }
+
+    output = writer.write_chunk(content="recognized text", metadata=metadata)
+
+    assert output.name == "SCR-001_0123456789ab.md"
+    text = output.read_text(encoding="utf-8")
+    assert "# Screenshot: SCR-001.png" in text
+    assert "recognized text" in text
+    assert "source_hash: 0123456789abcdef" in text