The test dataset in CT-AMAR is malformed

Some files in this dataset are output extract, not input files, and this leads to wrong data in the extract (`other id`-like data in the `Code genre d'activité` column).
The 3rd file (3-UPDATES.zip) is correct, but second file (1-CREATES.zip) file

Compare the files in the test dataset, especially 1-CREATES.zip and 3-UPDATES.zip, to determine which are input files and which are output extracts. Verify the generated extract's "Code genre d'activité" column against the corrected dataset; done means the malformed files are fixed and the incorrect "other id"-like data no longer appears.

Written by the indexing model from the issue text.