test(e2e): Playwright Smoketests gegen kochwas-dev (remote)

Automatisierte End-to-End-Tests gegen ein deployed Environment. Loest
die manuellen MCP-Playwright-Runs ab. 42 Tests in 9 Files:

- homepage: H1, Sektionen, Sort-Tabs, Console-Errors
- search: lokaler Treffer, Web-Fallback, Empty-State, Deep-Link
- profile: Switcher, Auswahl-Persistenz, Favoriten-Section, Guard-Dialog
- recipe-detail: Header, Portionen-Scaling (4->6), Favorit-Toggle,
  Rating-Persistenz ueber Reload, Gekocht-Counter, Wunschliste-Toggle
- comments: eigenen erstellen+loeschen via UI, fremder hat kein Delete
- wishlist: Seite, Sort-Tabs, Badge-Sync, requireProfile-Custom-Message
- preview: Guard ohne ?url=, echte URL parst, unparsbare zeigt error-box
- admin: alle 4 Subrouten + /admin redirect
- api-errors: parsePositiveIntParam (4x Invalid id), validateBody (4x
  Invalid body + issues), 404, Sanity /health /profiles /domains

Architektur:
- Separate playwright.remote.config.ts (getrennt von local preview)
- workers: 1 + afterEach API-Cleanup (rating, favorite, wishlist, comments)
- Hardcoded Recipe-ID 66 + Profile 1/2/3 — stabile Dev-DB-Seeds
- E2E_REMOTE_URL ueberschreibt die Ziel-URL

Ausfuehrung: npm run test:e2e:remote

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

playwright.remote.config.ts

@@ -0,0 +1,30 @@
+import { defineConfig } from '@playwright/test';
+
+// Zweite Playwright-Config fuer E2E-Smoketests gegen ein deployed
+// Environment (standardmaessig kochwas-dev.siegeln.net).
+//
+// Getrennt von playwright.config.ts, weil diese Tests:
+// - keinen lokalen Preview-Server starten
+// - gegen eine echte Datenbank laufen (daher workers: 1, afterEach-Cleanup)
+// - Service-Worker-Lifecycle nicht manipulieren (das macht offline.spec.ts lokal)
+//
+// Ausfuehrung: npm run test:e2e:remote
+// Ziel-URL ueberschreiben: E2E_REMOTE_URL=https://... npm run test:e2e:remote
+const BASE_URL = process.env.E2E_REMOTE_URL ?? 'https://kochwas-dev.siegeln.net';
+
+export default defineConfig({
+  testDir: 'tests/e2e/remote',
+  fullyParallel: false,
+  workers: 1,
+  retries: 0,
+  reporter: [['list'], ['html', { open: 'never', outputFolder: 'playwright-report-remote' }]],
+  use: {
+    baseURL: BASE_URL,
+    headless: true,
+    trace: 'retain-on-failure',
+    screenshot: 'only-on-failure',
+    // Service-Worker zulassen, aber keine Offline-Manipulation — die
+    // Tests hier pruefen Live-Verhalten gegen den Server.
+    serviceWorkers: 'allow'
+  }
+});