The S3 import worker fails on PDFs because it tries to load a JSON file that was not included in the package. This was fixed in arkindex/workers/import/s3!10 (merged)
