fix(dashboard): surface analytics overview timeouts instead of empty stats

[santhiprakash]

Summary

Raise the /analytics/overview client timeout and show a retryable error in Overview/Monitoring when the fetch fails, instead of silently rendering 0/N/A empty states.

Motivation

Fixes #396. High-traffic projects can exceed the dashboard API client's 15s default while aggregating analytics. The abort was caught by useAnalyticsData but OverviewTab/MonitoringTab ignored error, so users saw zeros and "No traffic data yet" even though the endpoint returns real data when called directly.

Related issue

Closes #396

Changes

• apps/dashboard/src/hooks/useProjectEndpoints.ts — use a 60s timeout for fetchOverview (same order of magnitude as other slow dashboard calls like deploy trigger).
• apps/dashboard/src/app/(dashboard)/projects/[id]/components/OverviewTab.tsx — render error card with retry (invalidateProjectCaches) when analytics.error is set.
• apps/dashboard/src/app/(dashboard)/projects/[id]/components/MonitoringTab.tsx — same error surface.
• apps/dashboard/src/i18n/locales/*/projects.json — analytics.loadFailed string in all locales.

Verification

python3 ../scripts/preflight_ship.py --repo oblien/openship --local . --branch main \
  --file apps/dashboard/src/hooks/useProjectEndpoints.ts \
  --must-contain 'endpoints.analytics.overview' \
  --must-not-contain 'ANALYTICS_OVERVIEW_TIMEOUT_MS' --issue 396
# PREFLIGHT CLEAR

cd apps/dashboard && bun run test src/i18n/i18n-parity.test.ts
# 2 passed

Notes / Risks

• Global network-error-handler still toasts on abort; this adds an inline, tab-local error so the empty-state widgets are not misleading.
• 60s is a pragmatic ceiling; if aggregation routinely exceeds that, a follow-up could add server-side caching rather than raising the client timeout further.

Checklist

• One change per PR — one bug, or one agreed feature, with nothing unrelated bundled in
• The diff is scoped — no reformatting or lint fixes on lines I wasn't otherwise changing
• A test fails without this change and passes with it (or I explained above why there isn't one)
• bun run test, bun run --cwd <workspace> lint, and bun format all pass locally
• I understand every line of this diff and can explain it in review

[Hydralerne]

MonitoringTab blanks the whole tab (should-fix)
The MonitoringTab change is a full early return (MonitoringTab.tsx:309) that replaces the entire MonitoringView with the error card when /analytics/overview fails. But that view renders several independent data sources:

geo map + top paths → useAnalyticsGeo (separate cache)
CPU/memory resource charts → useProjectUsageHistory (a DB read, not edge aggregation)
live traffic feed → useProjectUsageStream (SSE, already has its own usageError/reconnect)
So an overview timeout now hides the resource charts and live feed that loaded fine. That's a resilience regression: before this PR, an overview failure left analytics null and the TrafficChart showed its empty state, but geo/usage/live still rendered. The fix should scope the error to the traffic chart/summary block — exactly the way OverviewTab already scopes it — not nuke the tab.

[santhiprakash]

Pushed a follow-up commit to address the feedback.

The early return in MonitoringTab is gone. Instead, MonitoringView now accepts an analyticsError prop and scopes the retry card to the stats strip / traffic chart block only. The resources card, visitor map, live hits, top paths, response mix, and usage history all keep rendering when /analytics/overview fails, which matches the OverviewTab pattern.

What changed:

• Removed the whole-tab if (showAnalyticsError) early return from MonitoringTab.
• Added analyticsError and onRetryAnalytics props to MonitoringView.
• When analyticsError is set and not loading, MonitoringView replaces the stats strip and CollapsibleCard traffic block with the existing error card + retry button.
• The no-data prompt and traffic CollapsibleCard are suppressed while the error is shown.

Local verification:

• bun run --cwd apps/dashboard lint (tsc --noEmit) passes.
• bun run --cwd apps/dashboard test passes (447 tests, including MonitoringView.test.tsx).
• bunx prettier --write applied to the two modified files.

Please take another look when you have a moment.