Resume sending E2E test failures to slack

[deepakpathania]

Fixes #3555

Changes proposed in this Pull Request

• Add the required config for using e2e-environment to resume sending E2E test failures to slack.
• Rename jest-setup.js to jest.setup.js and move it to tests/e2e/config for it to be picked up correctly.
• Use useE2EJestConfig for jest config to use the failure script that includes slack reporting.

Testing instructions

• Go to Actions tab -> WC Pay E2E tests.
• Run the workflow for this branch update/slack-e2e-reporting.
• Notice failure messages would be sent to #wcpay-bots channel in slack, reference job.

Sample slack screenshot

---

• Added changelog entries to both readme.txt and changelog.txt (or does not apply)
• Covered with tests (or have a good reason not to test in description ☝️)
• Tested on mobile (or does not apply)

Post merge

• Added testing instructions to the release testing doc (or does not apply)

[naman03malhotra]

LGTM, left a comment for my understanding.

[zmaglica]

LGTM.

[dechov]

Was there a change to where the screenshots get saved? It happens here but maybe when running locally it depends on where the tests are run from? Not seeing the artifact on recent runs (e.g. here).

[deepakpathania]

@dechov We moved to use the useE2EJestConfig provided by e2e-environment which sends the failed screenshot messages directly to #wcpay-bots channel in slack, which might have overridden the old artifact behavior to only create artifacts for failure cases, see this run for reference.
Do we need the screenshots for success scenarios as well?

[dechov]

Ah, I see 👍

Not that we need the regular ones but there is some theoretical benefit to having snapshots from each distinct screen from the normal run, to ensure that there aren't visual regressions that wouldn't be caught by specific assertions. What do you think of having a "failure screenshots" artifact in addition to the earlier one?

[deepakpathania]

I feel unless there is an alert/checking process in place, we wouldn't really monitor the successful ones regularly, and manually checking visual regressions could be tricky across runs.
We could either ask GS to do some sanity or have explicit visual regression tests in place for the same but that's just my opinion, we could discuss it internally to see what other folks feel about the same.