bug: PDF parsing is incorrect

[DRMPN]

Description

Parsing this pdf https://arxiv.org/pdf/2504.11688 results in when validating doc or paper:

{'abstract': '', 'experiments': '', 'results': ''}

The reason is that some content form PDF is missing or not being found during parsing process.

In the debug logs it can be seen, that abstract section is missing before the request to LLM to find it is made.