WHY are these changes introduced?

E2E tests currently share a single pre-existing dev store configured via E2E_STORE_FQDN. This causes:

• Shared state: earlier test failures can leave installed apps or modified config that affect later tests
• No isolation: tests can't run independently — one test's side effects leak into the next
• Fragile teardown: the old teardownApp silently skips steps, doesn't verify success, and a single failure blocks later cleanup

This PR also lays the foundation for parallel workers (future PR), where shared stores would be impossible since multiple tests run simultaneously.

WHAT is this pull request doing?

Per-test store isolation:

• Tests that need a dev store (app dev, hot reload, multi-config dev, toml-config dev) create a fresh store via browser automation on admin.shopify.com/store-create/organization/{orgId}
• Each test gets its own isolated store — no shared state between tests
• Store is created in test setup and cleaned up in teardown

Robust teardown with escalating retry (setup/teardown.ts):

• 3 independent phases: uninstall app → delete store → delete app
• Each phase verifies success (e.g., "Add apps to your store" empty state, "Your plan was canceled" page)
• Escalating retry: retry same step → go back one step and retry both
• 90s timeout prevents teardown from blocking the suite
• Failure summary tells you what wasn't cleaned up and to run pnpm test:e2e-cleanup
• Phase 3 (delete app) always runs even if store cleanup failed

Tagged logging (setup/env.ts):

• createLogger(tag) produces [e2e][w0][browser], [e2e][w0][cli] format
• e2eSection() prints section headers: ----- SETUP: store ... -----, ----- TEARDOWN: app ... -----
• Worker index in all log lines for future parallel debugging

Server error handling (setup/browser.ts):

• refreshIfPageError(page) detects 500/502 error pages and auto-refreshes
• Used in dashboard navigation and app deletion retry loops

Files changed:

How to test your changes?

# Run all tests with debug output
DEBUG=1 pnpm --filter e2e exec playwright test

# Run headed to watch store creation/teardown
E2E_HEADED=1 DEBUG=1 pnpm --filter e2e exec playwright test

# Skip teardown for debugging (clean up manually after)
E2E_SKIP_CLEANUP=1 pnpm --filter e2e exec playwright test

# Verify no leftovers after a run
npx tsx packages/e2e/scripts/cleanup-stores.ts --list
npx tsx packages/e2e/scripts/cleanup-apps.ts --list

Checklist

• I've considered possible cross-platform impacts (Mac, Linux, Windows)
• I've considered possible documentation changes
• I've considered analytics changes to measure impact
• The change is user-facing, so I've added a changelog entry with pnpm changeset add