Support real language identification and fix test related to this

[lpi-tn]

This pull request introduces updates to the welearn_datastack OpenAlex plugin and its associated test suite. The changes focus on improving the accuracy of language detection for document content and refining the test cases to align with the updated functionality.

Core functionality updates:

• Introduced the langdetect library to dynamically detect the language of document content based on the extracted PDF text, replacing the previous reliance on the language field from the JSON response. (welearn_datastack/plugins/rest_requesters/open_alex.py) [1] [2] [3]

Test suite updates:

• Updated test cases to use realistic document content ("The findings highlight...") instead of placeholder text ("Lorem ipsum") for better alignment with production scenarios. (tests/document_collector_hub/plugins_test/test_open_alex.py) [1] [2]
• Removed assertions for deprecated fields (duration and readability) in the test cases, reflecting changes in the OpenAlex plugin's functionality. (tests/document_collector_hub/plugins_test/test_open_alex.py)