test(e2e): add Initiative 4 end-to-end + integration specs

Sales-process coverage (launch-readiness Initiative 4):
- exhaustive: full 7-stage sales journey + illegal-skip rejection + deposit
  total + tenancy/berth-sold; multi-berth EOI berth-range; EOI pathway parity
  (in-app vs Documenso, shared EoiContext); mobile-viewport journey.
- realapi (Documenso-gated, opt-in): generate-and-sign + post-EOI stages.
- integration: Documenso DOCUMENT_COMPLETED webhook idempotency (3x replay ->
  single file/audit write); storage backend swap (s3 <-> filesystem) with a
  real on-disk filesystem round-trip.
- visual: Reports UI snapshot cases (baselines captured separately).

1615 unit/integration pass; tsc + lint clean. Test-only change (specs are not
bundled into the app image) - no app behavior modified.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

tests/e2e/visual/reports.spec.ts

@@ -0,0 +1,69 @@
+import { test, expect, type Page } from '@playwright/test';
+
+import { login, navigateTo } from '../smoke/helpers';
+
+/**
+ * Initiative 4 (e2e testing) — visual regression baseline for the new
+ * Reports UI (Initiative 1, `docs/reports-content-spec.md`). Mirrors the
+ * list/landing pattern in `snapshots.spec.ts` so the two specs share the
+ * same project config (project=visual, matched by /visual/.*\.spec\.ts/).
+ *
+ * NOTE — NO baseline images are committed with this change. A human must run
+ *
+ *   pnpm exec playwright test --project=visual --update-snapshots
+ *
+ * once the Reports UI is visually settled to generate the PNGs under
+ * tests/e2e/visual/reports.spec.ts-snapshots/. Until then these cases will
+ * report "missing snapshot" on the first run (expected) — this is by design:
+ * the task brief explicitly asked NOT to generate/commit baselines, only to
+ * wire up the spec cases so the baselines can be captured later.
+ *
+ * The Reports landing (`/reports`) is a stable list/landing screen — it does
+ * not depend on per-row fixture data, so it tolerates seed drift the same way
+ * the other visual baselines do. The category sub-pages (sales / financial /
+ * operational) are charts-heavy; they're included but a generous
+ * maxDiffPixelRatio absorbs antialiasing differences across machines.
+ */
+
+const REPORT_PAGES = [
+  { name: 'reports-landing', path: '/reports' },
+  { name: 'reports-sales', path: '/reports/sales' },
+  { name: 'reports-financial', path: '/reports/financial' },
+  { name: 'reports-operational', path: '/reports/operational' },
+] as const;
+
+async function settle(page: Page) {
+  // Quiet animations / timers / blinking carets so dynamic content doesn't
+  // produce flaky pixel diffs (identical to snapshots.spec.ts).
+  await page.addStyleTag({
+    content: `
+      *, *::before, *::after {
+        animation-duration: 0s !important;
+        animation-delay: 0s !important;
+        transition-duration: 0s !important;
+        transition-delay: 0s !important;
+        caret-color: transparent !important;
+      }
+    `,
+  });
+  await page.waitForLoadState('networkidle');
+  // Let TanStack Query + chart libraries flush their first paint.
+  await page.waitForTimeout(750);
+}
+
+test.describe('Visual regression — Reports UI', () => {
+  for (const p of REPORT_PAGES) {
+    test(`${p.name} matches baseline`, async ({ page }) => {
+      await login(page, 'super_admin');
+      await navigateTo(page, p.path);
+      await settle(page);
+
+      await expect(page).toHaveScreenshot(`${p.name}.png`, {
+        fullPage: true,
+        // Charts/SVG render with sub-pixel antialiasing variance across
+        // machines — match the tolerance the existing visual baselines use.
+        maxDiffPixelRatio: 0.02,
+      });
+    });
+  }
+});