e2e-testing

# E2E Testing Patterns Reuse the project's existing E2E runner and conventions. New projects default to Playwright. One test should prove one user outcome through the same controls and feedback a user sees. ## Behavior Contract - Wait for the required heading, control, URL, status, or result; network silence and elapsed time are not readiness signals. - Prefer role, label, and visible text locators. Use a test ID only when no stable user-facing locator exists. - Keep setup deterministic and local. Reuse project fixtures, auth helpers, and data factories before adding another layer. - Assert the outcome in the test, including loading, empty, error, permission, recovery, and destructive states when they matter. ```typescript import { expect, test } from '@playwright/test' test('filters items', async ({ page }) => { await page.goto('/items') await expect(page.getByRole('heading', { name: 'Items' })).toBeVisible() await page.getByLabel('Search items').fill('alpha') await expect(page.getByRole('article').first()).toContainText(/alpha/i) }) ``` ## Extraction Boundary Use a page object only when several tests repeat the same stable navigation or interaction. Keep outcome assertions in the test; avoid generic base pages, wrappers for single locators, and a second fixture system. Organize tests by user capability using the repository's current layout. Do not reorganize an established suite merely to match a template. ## Coverage And Configuration Use a risk-based bro...