cms: 308-redirect old slugs to canonical URL on detail pages

[xrendan]

Summary

york_factory PR #43 ships editable slugs with friendly_id's :history module preserved — old slugs continue to resolve to the current canonical record on the CMS side. This wires up the consumer site to honor that:

• On every slug-based detail page (/memos/:slug, /toronto/memos/:slug, /posts/:slug, /builders/:slug), compare the requested slug against slug returned by the API. If they differ, permanentRedirect() (HTTP 308) to the canonical URL.
• Point alternates.canonical at the canonical slug everywhere (was using the URL param), and add it to /posts/:slug, which was missing one.

The resulting flow when an editor renames foo → bar:

1. /memos/foo hits york_factory, friendly_id resolves via history, returns the record with slug: "bar".
2. TradingPost sees the mismatch, returns 308 to /memos/bar.
3. Browser follows; /memos/bar returns 200 with <link rel="canonical" href="/memos/bar">.

Why 308 and not 307

Permanent (308) is the right call here — transfers link equity to the new URL and gets search engines to update the index. The cache-staleness trade-off (browsers cache 308s) is real but self-healing: worst case is one extra round trip on a re-rename. Discussed inline; revisit with Cache-Control: no-store middleware if editors start churning slugs frequently.

Test plan

• Rename a memo's slug in the york_factory admin; visit the OLD URL on TradingPost; confirm 308 → new URL in network tab.
• Repeat for a post, a builder, and a toronto memo.
• Visit the current canonical URL; confirm 200 (no redirect loop).
• Visit a never-existed slug; confirm 404 (not redirect).
• View source on a detail page; confirm <link rel="canonical"> points at the canonical slug.

🤖 Generated with Claude Code