Implement in-memory caching for projects.json

[Akshita-2307]

Summary

• Make load_all_projects() actually use the _projects_cache instead of reading from disk every call
• Cache is populated on first call and reused until clear_cache() is invoked (fixes [Bug]: Fix recommendation input validation and remove unused cache logic #673, [Bug]: [Performance] Implement In-Memory Caching for projects.json #472)

[vercel]

Someone is attempting to deploy a commit to the komalsony234-1530's projects Team on Vercel.

A member of the Team first needs to authorize it.

[Akshita-2307]

Program: GSSoC

[komalharshita]

@Akshita-2307 Thank you for the contribution.

I reviewed the changes and found that this implementation duplicates caching functionality that has already been proposed and integrated through earlier pull requests affecting the same data_loader.py logic.

The current PR introduces no additional functionality beyond the existing caching approach and also has merge conflicts with the current main branch.

Since the underlying issue has already been addressed and merging this PR would duplicate previously merged work, I am closing this pull request.

Thank you for your effort and contribution to the project. Please feel free to explore other open issues and submit new contributions.