ops: publish codeforphilly-data-snapshot public repo (referenced in architecture.md but doesn't exist)

## What's missing

\`specs/architecture.md\` and \`specs/behaviors/storage.md\` both reference a **public scrubbed snapshot** of the data repo — \`codeforphilly-data-snapshot\` — that contributors can clone without needing access to the live production data. The snapshot is supposed to be produced by \`apps/api/scripts/scrub-data.ts\` and pushed periodically.

The scrubber script exists. The public snapshot repo does not.

## Fix shape

Two pieces:

1. **One-time:** create the \`CodeForPhilly/codeforphilly-data-snapshot\` public repo on GitHub; do the first \`scrub-data\` run; push the result.
2. **Recurring:** add a scheduled CI workflow (GitHub Action on a weekly cron, say) that runs the scrub + force-pushes a fresh snapshot tag. Probably lives in \`.github/workflows/\` in this repo since the script lives here.

## Why now

Pre-cutover-onboarding stories assume new contributors can clone the snapshot. If we leave the spec asserting the repo exists without it existing, new contributors hit a dead link.

Identified during the 2026-05-30 post-cutover-blog spec-drift audit.

Start with apps/api/scripts/scrub-data.ts and read the references in specs/architecture.md and specs/behaviors/storage.md. Create the CodeForPhilly/codeforphilly-data-snapshot public repository, run the scrubber and push its result, then add a scheduled workflow under .github/workflows/. Done means the snapshot is public, initially populated, and refreshed weekly with a snapshot tag.

Written by the indexing model from the issue text.