don't fail on unreadable files in raw processing pipeline
A user accidentally included a __MACOSX
folder in the zip upload. This folder contains a bunch of borked files with internal metadata and a filename that matches the filename of an original upload.
Due to that the entire upload fails with an error Unexpected error: <zipfile.ZipExtFile name='__MACOSX/<REMOVED>/._4-2.pdf' mode='r' compress_type=deflate>: unable to find trailer dictionary while recovering damaged file
.
Instead of failing the complete thing, we should just skip the broken files.