Merge the Transkribus and file imports

• Remove arkindex_tasks.export_transkribus
• Move the Transkribus import code from arkindex_tasks.import_transkribus to arkindex_tasks.import_files.transkribus, ignoring the main()
• Move arkindex_tasks.pagexml to arkindex_tasks.import_files.pagexml
• Remove support for job paths from the TranskribusImporter
• Have the TranskribusImporter return its generated elements instead of dumping elements.json file by itself
• Add code in arkindex_tasks.import_files to:
  • Detect that a file is a ZIP archive
  • Try to open the ZIP archive
  • If the ZIP archive contains a mets.xml, assume that it is a Transkribus archive and import it using the TranskribusImporter
    (Yes, we do not support a generic METS import, but mets.xml is the only file that should always be found in every Transkribus export)
  • If it does not contain mets.xml, die in mysterious circumstances (for now)