How many file_meta still missing core metadata?

    SELECT COUNT(*) FROM file_meta WHERE sha256hex IS NULL;
    => 1,130,915

Great! Not many.

And are in petabox?

    SELECT COUNT(*)
    FROM file_meta
    LEFT JOIN petabox ON file_meta.sha1hex = petabox.sha1hex
    WHERE file_meta.sha256hex IS NULL
      AND file_meta.sha1hex IS NOT NULL;
    => 1,149,194

Almost all; maybe just some CDX fetch failures or something in there. So,
should run these on, eg, grobid2-vm.

    COPY (
      SELECT row_to_json(petabox.*)
      FROM file_meta
      LEFT JOIN petabox ON file_meta.sha1hex = petabox.sha1hex
      WHERE file_meta.sha256hex IS NULL
        AND file_meta.sha1hex IS NOT NULL
    ) TO '/grande/snapshots/dump_grobid_petabox_todo.json';

Count of PDF files that GROBID processed and matched to a release (via
glutton), but no PDF in `fatcat_file` (note: `fatcat_file` is out of date by a
couple million files):

    SELECT COUNT(*) as total_count, COUNT(DISTINCT grobid.fatcat_release) as release_count
    FROM grobid
    LEFT JOIN fatcat_file ON grobid.sha1hex = fatcat_file.sha1hex
    WHERE fatcat_file.sha1hex IS NULL
      AND grobid.fatcat_release IS NOT NULL;

     total_count |  count  
    -------------+---------
         5072452 | 4130405