docs: retire the tailwind pin, refresh the docs for the fifth proceeding

[leegitw]

Two doc commits. Nothing in src/ changes.

1. The Tailwind pin is retired, after checking rather than reverting

Dependabot #28 merged and moved tailwindcss + @tailwindcss/postcss from exact 4.3.0 to 4.3.3 — a range CLAUDE.md explicitly named as not resolving.

The upstream problem is fixed: clean npm ci resolves, lint and build pass, and a live contrast audit over 177 text nodes finds zero failures with the opacity colours still emitting. The pin is recorded as history so nobody re-applies it from a warning that no longer holds. typescript-eslint 8.59.4 stays pinned; that one is still real.

The near-miss is the part worth keeping. After the merge, package.json and the lockfile said 4.3.3 while node_modules/ still held 4.3.0. Every local build passed, against a version the repo no longer declares. Railway installs clean, so production was building something local had never compiled. New rule: after any dependency merge, run npm ci before trusting a green build. npm run build alone verifies the wrong tree, and it fails in the direction that looks like success.

2. Docs refreshed for the fifth proceeding and the corrected calendar

Four places still described the pre-Tennessee world:

• README listed three case files; there are four, and the fourth is the only one in trial.
• CLAUDE.md's EU DSA thread said /lawsuits "calls the record four proceedings". It says five.
• The SEO plan's news calendar pointed at a trial that no longer exists. The second California bellwether was its lead near-term trigger; Snap settled July 20 and the plaintiff dismissed Meta July 22. July 27 is still a date, for the Tennessee openings, with the federal AG trial August 12 and 18. The entry now carries the lesson: a scheduled trial is not a guaranteed news event.
• The ledger-candidates plan listed Tennessee only as "in trial now" under open questions. It now records something more useful: Tennessee was never in the section 4 candidate list. It surfaced from the July-25 status re-check, not from the nine research passes. A document enumerating candidates should say plainly when the thing that got built came from outside its own enumeration, or a reader treats the list as closed.

Checklist

• npm run lint and npm run build pass (against a clean npm ci tree)
• House style clean
• Paper body in sync with its review copy
• Every relative doc link resolves
• No sibling-repo references
• Docs-only; no runtime change

🤖 Generated with Claude Code