Test the whole buying journey and write a pre-release report

You are a careful senior test engineer working inside my project folder. The store is "Thread & Needle", a fashion e-commerce store selling women's and men's clothing, shoes and accessories in United States, charging in USD, with payments through Stripe at threadandneedle.com. Prove that it works and tell me honestly if it is not ready. I am not a programmer, so explain any technical word in a few words.

Plan first: read the cart, pricing, stock, checkout, order and webhook code, then show me a test plan (what you will test and what you will not) and wait for my OK before changing anything.

Tasks:
1. Unit tests with Vitest for the money and stock maths: line totals, cart totals, shipping rules, tax if the store charges it, discounts if any, rounding in the smallest unit of USD, refund limits, and stock never going below zero. Include edge cases: empty cart, quantity zero, huge quantities, two buyers for the last item. If you find a bug, show it to me, fix it in a separate commit and keep the test.
2. A separate test database called shop_test with migrations applied and a small seed. Tests must refuse to run if the database URL points anywhere except a local database with test in its name. Never touch production data.
3. End-to-end tests with Playwright for the full buying journey in test mode: open the home page, find a product, add it to the cart, change quantity, check out as a guest, reach the Stripe hosted page, then complete payment. Do not try to automate the gateway's hosted page if it blocks automation. Instead, create the order up to the redirect, then send a correctly signed test webhook to the local site, as the gateway would, and confirm the order becomes paid, stock drops once, and a second delivery of the same webhook changes nothing. Capture the order confirmation email through a test stub, not real sending.
4. More end-to-end tests: customer sign-up and login, a wrong password, an admin marking the order shipped, an admin refund, and a customer being refused access to /admin.
5. Accessibility checks with axe through Playwright on the home, product, cart, checkout, login and an admin page. Fix serious and critical problems such as missing labels, low contrast and keyboard traps, and list the rest.
6. Performance checks with Lighthouse on the home, product and cart pages, on a production build run locally. Report the scores and the main causes of slowness (large images, scripts). Fix the easy wins, such as image sizes and missing dimensions, and list the rest. Do not invent target numbers; report what you measure.
7. Add npm scripts: test, test:e2e and a single check script that runs type-check, lint, unit tests and a production build.

Security rules: never print, log or commit secrets; use only test keys; keep test data fake.

Verify: run everything and show me the real output, including failures. Do not hide or skip failing tests to look green.

Then write TEST_REPORT.md with: what was tested, results, bugs found and fixed, known gaps, and a clear ready or not ready verdict for release with reasons.

Commit on a feature branch with a clear message.

Report back in this format: 1) What was done. 2) What I should check myself. 3) Anything that needs my decision.

Review the prompt, then copy it. Never paste real passwords or API keys into a chat.