AI search feature

[nshaff3r]

Description of Change

Users can now search the library with AI by simply asking a question in the search box!

Related Issue

Motivation and Context

Features:

• Returns a concise summary along with relevant documents, creating a more powerful search than keywords.
• Fetches entire document base for AI query using existing infrastructure that supports caching.
• Supports chunking of AI requests for models with smaller token limits.
• Can query multiple times and combine responses for greater accuracy using the LLM_API_CALLS environment variable.
• Makes API and fetch requests in parallel.
• Defaults to a null answer to avoid hallucinations.
• Uses both desktop and mobile-friendly front-end animations.

Checklist

• Ran npm run lint and updated code style accordingly in cache.js and docs.js I replaced requireWithFallback with require to get my vercel deployment to work. This will need to be undone.
• npm run test passes The only failing tests are due to my disabling user auth to get the demo site to work.
• PR has a description and all contributors/stakeholder are noted/cc'ed
• tests are updated and/or added to cover new code
• relevant documentation is changed and/or added