PyPDF performs badly on many types of PDF documents.
We add a text extraction with pdfminer, which is designed for this task.
Because pdf content extraction was so flaky, it was completely
deactivated by 1b753b0d53. We revert that :-)
closesodoo/odoo#38508
Task: 2152494
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
