Make self-supervised OCR config portable

[sushantkhemalapure]

Summary

This PR makes the self-supervised OCR subproject easier to run on different machines by removing machine-specific config defaults.

Changes

• replace absolute /home/... dataset paths in RenAIssance_SelfSupervisedLearning_OCR_YukinoriYamamoto/config.json with repo-relative paths
• move default saved model outputs to the local models/ directory
• add check_config_paths.py to validate configured dataset/model paths before running notebooks
• add portable data/ and models/ scaffolding for expected local layout
• update the self-supervised README with setup instructions for the new relative-path structure
• update .gitignore to avoid committing local datasets and checkpoint files from this subproject

Why

The previous config only worked on one specific local filesystem layout. With relative defaults and a simple path checker, contributors can clone the repo and adapt the setup more easily.

Testing

• ran python check_config_paths.py
• verified the new default relative paths resolve correctly in the self-supervised OCR directory

Notes

• placeholder folders/files were added only to document the expected structure
• no real datasets or model weights are included in this PR