feat(templates): add openai-vector-store-search template

[jacobdanner]

Description

Adds a new openai-vector-store-search Function template that queries an OpenAI Vector Store for the documents most relevant to a search query, using the official openai Node SDK.

• /search Function — accepts query (required) plus optional vector_store_id, max_num_results, filters, score_threshold, and rewrite_query, and returns the raw OpenAI search response. Falls back to an OPENAI_VECTOR_STORE_ID env var when a request doesn't supply its own vector_store_id, and trims whitespace from either source before use (a stray leading/trailing space in an env var otherwise produces a confusing 400 from OpenAI).
• Self-contained package — the template's own package.json declares all of its dependencies/devDependencies and its own test/lint/format scripts, ESLint/Prettier config, and ignore files, so it can be copied out of this workspace and run standalone (npm install && npm test with no access to the repo root).
• Tests — an SDK-mocked suite (tests/search.test.js, 12 cases) plus a small nock-based suite (tests/search.endpoint.test.js) that lets the real SDK build and send its HTTP request and intercepts it at the network layer, to catch drift between what we assume the SDK sends and what it actually sends.
• scripts/smoke-test.js (npm run smoke-test) — hits a real deployed /search endpoint via FNA_SERVERLESS_EMBEDDINGS_ENDPOINT to verify a deployment works end-to-end, not just against local mocks.
• Also fixes a path bug in _helpers/template-docs.js that broke automatic regeneration of docs/templates.md (it wrote to a path relative to process.cwd() instead of __dirname), and regenerates that file — which is why the diff includes updates to several unrelated templates' listings.

A note for reviewers

Pre-existing, unrelated to this change: npm run lint and npm test at the repo root currently fail even on main —

• international-telephone-input/functions/lookup.js has 3 multiline-comment-style violations that predate this branch (confirmed identical to origin/main), which trips the whole-repo npm run lint (and the pre-push hook, and this repo's nodejs.yml CI).
• 10 unrelated test suites (video-token, voice-client-javascript, sync-token, chat-token, google-sheets, covid-vaccine-faq-bot, etc.) fail under the current Node version, unrelated to this PR.

openai-vector-store-search's own tests and lint are clean in isolation (npx eslint --ext js openai-vector-store-search, npx jest openai-vector-store-search — 0 errors, 12/12 passing). Flagging so the repo-wide CI failure isn't mistaken for something this PR introduced.

Checklist

• I ran npm test locally and it passed without errors.
• I acknowledge that all my contributions will be made under the project's license.

Related issues