Canvas - skip TutorProblemFile OCR if content is the same

[shanbady]

What are the relevant tickets?

Closes https://github.com/mitodl/hq/issues/8766

Description (What does it do?)

This PR resolves an issue where we run OCR for problem file pdfs even if the content hasnt changed which means on every webhook sync sent by the data platform (for canvas courses that have pdf problem files)

How can this be tested?

1. checkout this branch
2. in settings_course_etl.py set CANVAS_PDF_TRANSCRIPTION_MODEL to an ollama model that can process images (in my case phi4-mini) or gpt-4o-mini if you are setup with openai
3. restart your celery container
4. run a sync for a canvas course that has problem files - keep an eye on the celery logs and you should see output indicating that the task is calling an llm to ocr files: python manage.py backpopulate_canvas_courses --canvas-ids 14566 --overwrite
5. re-run once again and note that the LLM is not called for processing the problem files

[dsubak]

LGTM