Currently CI fails on master. This PR is an attempt to clean up CI so it's all green.

It coveres two things:

1. Make the tests more stable: It turned out that it was mostly a race-condition issue, and increasing the wait times at certain critical locations turned out to do the trick. This is obviously not a very nice solution, but it's the solution that the current test-suite employs, so without rewriting everything, this is what I came up with.
2. Switch from Travis to GitHub Actions: Over at the hapi GitHub org, they recently switched to using GitHub Actions for CI instead of Travis. There is even a PR to do the same with @hapi/good (Create ci-plugin.yml hapijs/good#639). However, that PR hasn't yet been merged because their CI is also flaky just as ours is. I'm planning to contribute the changes I came up with here to them I contributed my findings upstream (Fix CI and switch to GitHub Actions hapijs/good#640).

Note that the added .github/workflows/ci-plugin.yml file is identical to the one in hapijs#639 with one notable exception: Ours also tests against hapi v18, since we're currently using that in Kibana.