Clay embeddings on Sentinel-1 data

[anindabitm]

Added notebook to demonstrate use of Clay embeddings on Sentinel-1 imagery for earthquake detection

[srmsoumya]

Thank you, @anindabitm, for the well-formatted and clean notebook.

Before we proceed with merging the pull request, here are a few additions I recommend:

• Please include a brief description of the dataset and the problem statement to give context to anyone reviewing this work.
• It would be beneficial to elaborate on our approach using the CLAY model. Specifically, explain how we're utilizing CLAY embeddings for analyzing pre/post-earthquake scenarios and the rationale behind building a classifier on top of these embeddings.
• Could you switch to using the actual validation and test sets? This change will help in providing a more accurate evaluation of the embeddings' effectiveness and their performance relative to the leaderboard.
• Lastly, some insights into the time required to generate these embeddings and to train a robust model using them would be valuable for understanding the computational efficiency and feasibility of our approach.

Looking forward to the updates, and thank you for all your hard work.

[anindabitm]

Thanks @srmsoumya for your inputs. I have updated the notebook. Request you to please review

[lauracchen]

@yellowcap is this ready to be merged?