Meme Search v2.3.2: refreshed dependencies and cleaner CI

This maintenance release refreshes the image-to-text runtime, security-related
Ruby dependencies, frontend test tooling, and GitHub Actions. It also fixes the
image-to-text lint job so Ruff checks the intended component with its checked-in
configuration instead of scanning unrelated Python files across the repository.

What changed

• The image-to-text service now requires PyTorch 2.13 or newer, bitsandbytes
  0.50 or newer, and psutil 7.2.2 or newer. The combined dependency set passed
  the service's lint and pytest checks before release.
• The Rails application now uses Ruby OpenSSL 4.0.2. SimpleCov and the frontend
  test dependencies were also refreshed.
• GitHub Actions now uses actions/setup-python@v7 across the affected test and
  container-build workflows.
• The Ruff action now receives its source path and configuration through inputs
  supported by chartboost/ruff-action@v1.

Upgrade

Back up your database, then pull and recreate the containers:

docker compose pull
docker compose up -d
docker compose exec meme_search bin/rails db:migrate:status

No new database migration or Search API contract change is included. Existing
API tokens, clients, indexed libraries, and Compose configuration remain
compatible. Source installations of the image-to-text service should reinstall
its Python requirements before restarting it.

Search API security boundary

Search API v1 remains permanently read-only and officially supported only on
loopback. Its bearer tokens do not authenticate the web UI or settings. Do not
expose Meme Search directly to a public network.

• Search API guide
• Watch the 16-second app-and-CLI demo
• Propose a community integration