fix: align homepage coverage and freshness semantics

[jerdaw]

Summary

• server-render exact national coverage from verified, visible database rows
• replace the static four-hour freshness claim with an hourly source-check contract and explicit scheduler/upstream caveats
• clarify that /api/status grades distinct UTC measurement-hour completeness while /api/health reports current heartbeat freshness
• extend production smoke coverage for SSR counts, legacy placeholders, cadence copy, and status-basis metadata

Production evidence

At 2026-07-22T02:24:54Z, all four source heartbeats were healthy at about 88 minutes, while each source had 11 of 24 distinct measurement hours. This correctly produced /api/health.healthy=true and /api/status.system_uptime_24h=0.458; recent GitHub scheduled runs were successful but irregularly spaced.

Verification

• focused frontend tests: 28 passed
• frontend type-check, lint, format check: passed
• production build: passed
• documentation/roadmap gate: passed
• backend documentation tests: 15 passed
• Playwright desktop hydration/refresh and Mobile Chrome coverage checks: passed in the prior local verification

Known prerequisite caveats

• the broader frontend suite retains five existing equity tests that require the absent Ontario GeoJSON asset
• type-check:test retains older unrelated Vitest mock-generic errors
• Mobile Safari remains unavailable locally because the WebKit binary is not installed

Deployment boundary

The private service inventory marks this frontend deployment manual-only. The current checkout does not contain the historical workstation release helper, so this PR does not improvise or execute a production deploy.

[jerdaw]

Post-merge production snapshot

Canonical main commit: 744fd5c6c86bc81db0023c928383f080fd152a41

API/HTML check at 2026-07-22T02:54:03Z:

• /api/hospitals.count=399
• /api/hospitals.coverage=null (old deployed API)
• /api/health.healthy=true, stale threshold 120
• /api/status.overall_status=critical, system_uptime_24h=0.458
• /api/status.overall_status_basis=null (old deployed API)
• HTML still contains ...+ Hospitals and Fresh Data Every 4 Hours
• HTML does not contain Sources Checked Hourly

Independent Chromium check at 2026-07-22T02:56:08Z:

• SSR/no JavaScript: 4 Provinces...+ HospitalsFresh Data Every 4 Hours
• after hydration: 4 Provinces / 223+ Hospitals / Fresh Data Every 4 Hours
• after hard refresh: same hydrated result
• mobile horizontal overflow: false

The 223 hydrated value is the Ontario-filtered list length, not the national API count of 399.

Production Smoke run 29887041443 failed as expected before deployment:

• FAIL homepage coverage - missing hospital_count
• FAIL /api/status - missing expected content: "overall_status_basis":"measurement_hour_completeness_24h"

All other legacy route/API smoke checks passed. The private service inventory marks deployment manual-only, and the current checkout lacks the retained historical workstation release helper, so no deployment was improvised.