Search: replace DOM-based search with pre-computed search index

Blocked by

Issue "Refactor archive system: manifest + lazy loading + pre-computed contribution count" — implement only after that is stable.

Note

Future improvement. The lazy-loading refactor (above) solves full-archive search by fetching all files on search focus. This issue goes further: a pre-computed index makes search instant without downloading all archive JSON.

Problem

Even after the lazy-loading refactor, searching still means downloading all archive JSON (~10–13 MB, growing linearly). On slow mobile this is a poor experience.

Fix

At archive time, generate search-index.json alongside manifest.json:

[
  { "name": "Jane Doe", "about": "...", "handles": ["@jane"], "resources": ["title1", "title2"], "file": "archive_12.json" },
  ...
]

~200–400 KB for 4,500 cards — smaller than two archive files. Load eagerly on page open. Search runs against the index instantly. When a match points to an un-loaded archive file, fetch that specific file on demand to render the card.

Files

• scripts/card-to-archive.js — updated to also generate search-index.json
• archive/search-index.json — new (auto-generated)
• assets/script.js — updated: load index eagerly, search against index, fetch specific archive files on match

This work is blocked by the archive manifest and lazy-loading refactor, so confirm that dependency is stable first. Read scripts/card-to-archive.js and assets/script.js, then inspect archive/search-index.json generation and loading behavior. Done means the index is generated and loaded eagerly, searches run against it, and matching archive files are fetched on demand.