dev: Metabase cleanup

[isTravis]

The only remaining use of Metabase on the app was an old /admin page dashboard. This PR updates that dashboard to use native PG data, and removes all remaining metabase code.

The /admin page previously embedded a Metabase iframe that displayed the
"PubPub Top Metrics" dashboard. This has been replaced with a native Analytics
tab on the Superadmin dashboard (/superadmin/analytics). The old /admin URL
now 301-redirects to /superadmin/analytics.

All data is served by a single API endpoint (GET /api/platformAnalytics) that
is restricted to superadmin users. Results are cached in-memory for 1 hour.

A second endpoint (GET /api/platformAnalytics/period) provides date-range
filtered counts and a paginated new-communities table.

---

Files changed / created

File	Change
server/platformAnalytics/api.ts	New. API endpoint with two routes: all-time overview + period drill-down.
client/containers/SuperAdminDashboard/PlatformAnalytics.tsx	New. React component with recharts graphs and date-range explorer.
client/containers/SuperAdminDashboard/platformAnalytics.scss	New. Grid layout with responsive breakpoints.
client/containers/SuperAdminDashboard/tabs.tsx	Added analytics tab entry.
utils/superAdmin.ts	Added 'analytics' to superAdminTabKinds.
server/apiRoutes.ts	Registered platformAnalyticsRouter.
server/routes/adminDashboard.tsx	Replaced Metabase iframe SSR with a 301 redirect.
client/containers/index.ts	Removed AdminDashboard export.
client/containers/App/paths.ts	Removed AdminDashboard path entry.

---

Graph details

All-Time Totals (scalar cards)

Metric	Query	Source
Total Communities	SELECT COUNT(*) FROM "Communities"	Communities table
Registered Users	SELECT COUNT(*) FROM "Users"	Users table
Total Pubs	SELECT COUNT(*) FROM "Pubs"	Pubs table
Total Pageviews	SELECT SUM(page_views) FROM analytics_daily_summary	analytics_daily_summary materialized view (pre-aggregated from AnalyticsEvents)

Cumulative Growth (area charts)

Each chart shows total-to-date as an area fill. Data is computed client-side by
cumulatively summing the monthly series.

Chart	Query	Source
Communities	GROUP BY DATE_TRUNC('month', "createdAt")	Communities table
Users	GROUP BY DATE_TRUNC('month', "createdAt")	Users table
Pubs	GROUP BY DATE_TRUNC('month', "createdAt")	Pubs table
Pageviews	GROUP BY DATE_TRUNC('month', date) with SUM(page_views)	analytics_daily_summary materialized view

The server returns raw per-month counts. The client component runs a running
sum to produce the cumulative series for the area chart.

Monthly Growth (line charts)

Same underlying data as Cumulative Growth, but plotted as raw monthly counts
(no accumulation). Each line shows how many new entities were created that month.

Active Communities (dual line chart)

Two lines on the same chart:

Line	Query	Source
By Activity	COUNT(DISTINCT "communityId") GROUP BY month	ActivityItems table – counts distinct communities with any activity item per month
By Pub Creation	COUNT(DISTINCT "communityId") GROUP BY month from Pubs	Pubs table – counts distinct communities that had at least one pub created per month

Series are merged client-side by month key and sorted chronologically.

Period Explorer (bottom section)

Driven by GET /api/platformAnalytics/period?startDate=…&endDate=…&page=….

Metric	Query	Source
New Communities	COUNT(*) FROM "Communities" WHERE "createdAt" BETWEEN …	Communities table
New Users	COUNT(*) FROM "Users" WHERE "createdAt" BETWEEN …	Users table
New Pubs	COUNT(*) FROM "Pubs" WHERE "createdAt" BETWEEN …	Pubs table
Pageviews	SUM(page_views) FROM analytics_daily_summary WHERE date BETWEEN …	analytics_daily_summary materialized view
New Communities Table	SELECT title, subdomain, createdAt, description FROM "Communities" WHERE "createdAt" BETWEEN … ORDER BY "createdAt" DESC LIMIT 25 OFFSET …	Communities table

Preset buttons: 30 days, 90 days, 1 year, 2 years. Custom date inputs use
native <input type="date">.

---

Performance

• In-memory cache with 1-hour TTL for the all-time overview endpoint.
• Materialized views: Pageview total and monthly series read from
  analytics_daily_summary (pre-aggregated by community + day) instead of
  scanning millions of raw AnalyticsEvents rows.
• Period endpoint is not cached (date range varies) but queries are fast
  because they filter on indexed createdAt columns and the matview's
  (communityId, date) unique index.