feat(backend): clean up chunk PDFs on document and chunk deletion

Delete document endpoint now removes associated chunk PDF files from document_chunk/ before ChromaDB deletion. Delete chunk endpoint removes individual chunk PDF. Missing files logged as warnings, not errors.

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

backend/app/routers/documents.py

@@ -1,5 +1,6 @@
 """Documents CRUD router for RAG Database management."""
 import logging
+import os
 
 from fastapi import APIRouter, HTTPException
 
@@ -62,6 +63,16 @@ async def list_chunks(document_id: str):
     ]
 
 
+def _remove_chunk_file(file_path: str, base_dir: str) -> None:
+    """Remove a chunk file from disk. Logs warning if file is missing."""
+    full_path = os.path.join(base_dir, file_path)
+    try:
+        os.remove(full_path)
+        logger.info("Removed chunk file: %s", full_path)
+    except FileNotFoundError:
+        logger.warning("Chunk file not found, skipping: %s", full_path)
+
+
 @router.delete("/documents/{document_id}", response_model=DeleteResponse)
 async def delete_document(document_id: str):
     from app.core.config import get_settings
@@ -69,6 +80,15 @@ async def delete_document(document_id: str):
     settings = get_settings()
     rag = RAGService(settings=settings)
 
+    # Collect chunk file paths before ChromaDB deletion (metadata lost after delete)
+    chunks = rag.list_chunks(document_id)
+    chunk_paths = [
+        c["chunk_file_path"] for c in chunks if c.get("chunk_file_path")
+    ]
+
+    for rel_path in chunk_paths:
+        _remove_chunk_file(rel_path, settings.document_chunk_path)
+
     success, deleted_count = rag.delete_document(document_id)
 
     if not success:
@@ -89,6 +109,12 @@ async def delete_chunk(chunk_id: str):
     settings = get_settings()
     rag = RAGService(settings=settings)
 
+    # Get chunk metadata before ChromaDB deletion (metadata lost after delete)
+    result = rag.collection.get(ids=[chunk_id], include=["metadatas"])
+    metadatas = result.get("metadatas", [])
+    if metadatas and metadatas[0].get("chunk_file_path"):
+        _remove_chunk_file(metadatas[0]["chunk_file_path"], settings.document_chunk_path)
+
     success = rag.delete_chunk(chunk_id)
 
     if not success: