Add NodeNorm web frontends: GitHub Pages and Python-based

[gaurav]

This PR adds a web interface to babel-explorer, exposing all four tools — NodeNorm, XRefs, IDs, and Test Concordance — through a browser UI, a JSON REST API, and CSV downloads. It also includes the full CLI implementation, core query engine, and comprehensive test suite that the web frontend builds on.

WIP. TODO:

• Think about moving the Python library into a subdirectory (python/?) for better organization, or at the least excluding web/ from being packaged into a future PyPI package.

Why FastAPI?

We considered several options for the web framework:

• Streamlit — Easy to prototype with, but its re-run-the-whole-script model doesn't map well to babel-explorer's architecture (shared BabelDownloader/BabelXRefs instances with LRU caches, lazy multi-GB file downloads). It also wouldn't give us a clean REST API for programmatic access.
• Flask — A solid choice, but we'd need to bolt on separate API documentation (Swagger/OpenAPI) and pick additional libraries for request validation.
• Django — Too heavyweight for a tool with no database models or user auth.
• FastAPI was the best fit because:
  • It auto-generates Swagger docs at /docs, giving us a REST API with interactive documentation for free.
  • It runs synchronous route handlers in a threadpool automatically, which is exactly what we need — the core code uses synchronous requests and DuckDB, and some queries trigger multi-GB Parquet downloads that would block an async event loop.
  • Jinja2 templating is a first-class integration, so we get server-rendered HTML without a JS build system.
  • Combined with htmx, forms submit via AJAX and swap in HTML fragments without writing any frontend JavaScript framework code.

What's included

• Web app (src/babel_explorer/web/) — FastAPI app factory, routes, Jinja2 templates with Bootstrap 5 + htmx
• Four tool pages with forms that return results as HTML tables via htmx
• JSON REST API (/api/nodenorm, /api/xrefs, /api/ids, /api/test-concord) with query parameter interface
• CSV downloads for all tools (/api/*/csv)
• NodeNorm instance dropdown populated from NodeNorm.URLs with a "Custom URL..." option
• CLI command babel-explorer web with --host, --port, --reload options
• Navbar with links to all tools, Swagger docs, and GitHub repo
• 25 unit tests (tests/test_web.py) covering HTML pages, htmx partials, JSON API, CSV endpoints, and helper functions — all mocked, no network needed
• Core modules: BabelDownloader, BabelXRefs, NodeNorm, Click CLI, and 80+ tests across test_downloader.py, test_babel_xrefs.py, test_nodenorm.py

How to test

uv sync --group dev
uv run babel-explorer web          # http://127.0.0.1:8000
uv run pytest tests/test_web.py -v # 25 tests, ~0.4s