refactor: Remove pymupdf coupling in extraction

The library is only needed for pdf extraction which is taken care of
in its own extractor plugin. In the overall extraction routine we do not
need any knowledge of the existence of pymupdf.

papis_extract/extraction.py

@@ -2,13 +2,13 @@ import re
 from pathlib import Path
 from typing import Protocol
 
-import fitz
 import papis.config
 import papis.document
 import papis.logging
 from papis.document import Document
 
 from papis_extract.annotation import Annotation
+from papis_extract.extractors import ExtractionError
 
 logger = papis.logging.get_logger(__name__)
 
@@ -39,8 +39,8 @@ def start(
 
         try:
             annotations.extend(extractor.run(fname))
-        except fitz.FileDataError as e:
-            print(f"File structure errors for {file}.\n{e}")
+        except ExtractionError as e:
+            print(f"File extraction errors for {file}.\n{e}")
 
     if not file_available:
         # have to remove curlys or papis logger gets upset