Implement storage state auth optimization

[madrian13]

Optimizes Playwright test execution by implementing storage state to reuse authentication sessions across tests, eliminating redundant login operations.

Key Changes:

Add session-scoped authenticated_state fixture that saves browser auth state
Modify all page fixtures to reuse saved authentication
Add @pytest.mark.no_auth support for tests requiring fresh login
Add .auth/ directory to gitignore

Benefits:

~9% faster test execution (~2.5 min savings on full suite)
Reduced server load from fewer login requests
More reliable tests with less network dependency
Backward compatible - no breaking changes

[crivetimihai]

Thanks for this, @madrian13! Using Playwright's storage state for auth reuse is the right approach here.

A few observations:

1. Overly broad .gitignore - Adding *.json will ignore ALL JSON files in tests/playwright/. This could hide legitimate test fixtures or config. Consider narrowing to just:

   .auth/
   

   That already covers admin_state.json.

2. context fixture change - The original used new_context from pytest-playwright, which integrates with artifact collection (screenshots on failure, video recording). The new implementation creates contexts directly via browser.new_context(), which may bypass those hooks. Worth verifying screenshot-on-failure still works.

3. Repeated boilerplate - Every page fixture now has the same if no_auth: login else: goto+wait pattern, duplicated ~12 times. Consider extracting into a helper like _ensure_page_ready(page, base_url, request).

4. Documentation - storage-state-optimization.md is clear with before/after metrics. The 9% improvement should scale as more tests are added.

5. Edge case - The 1-hour TTL uses time.time() - stat().st_mtime. If the system clock shifts (NTP sync in CI), this could cause unexpected behavior. Minor, but worth noting.