Spike: Find better solution for Process.sleep() for FailureAlert tests

[taylordowns2000]

Context: Flaky tests are more expensive than slow tests.

What's up:

1. 689 introduces a 100ms sleep in two places to try to address a persistently flaky test.

2. Relying on Process.sleep() isn't great because it slows down our tests—in this case by 200ms per run.
3. But we think it's better than having this test flake once every 15 or 20 runs.

In other words, we've fixed this with Process.sleep for now to mitigate the impact of the flaky test on our devs, but there's gotta be a better solution.

[stuartc]

Pending outcome of https://github.com/OpenFn/Lightning/pull/689/commits/e78ebed653dc416ec1e0a1730196311c1b535955

[stuartc]

@taylordowns2000 https://app.circleci.com/pipelines/github/OpenFn/Lightning/2650/workflows/aab73471-ac61-4018-b75b-4f373a6c92ab/jobs/13073

Bruh.

Failed again.