In Odoo 16, the upload of the documents to the extraction server was moved to an asynchronous scheme through the usage of crons (see task 2888454).

While this looked like a good idea at first, it turned out that there were some issues in practice with this solution:
- Even when triggering the cron instantly, we have no guarentee that it will run right away as the cron workers could be busy with other tasks.
- Low feedback to the user on the state of the upload as it was happening in the background.

The main reason for this change was that it improved the reactivity of the interface as the upload itself occurred in the cron job.

Considering the situation, it has been decided that we'd go back to a synchronous scheme as the cons mentionned above outweighted the pros.

Some improvements have been made to mitigate the latency due to the re-introduction of synchronous uploads:
- On the OCR server side, the performances of the parsing routes have been improved to reduce its response time.
- An unnecessary call to get the IAP credits of the OCR service has been removed, this will be handled by the OCR server itself.

Task #3263472

Co-authored-by: Louis Baudoux <lba@odoo.com>