There were a ton of problems with these tests, but the main one was the use of `waitFor` without `await`, causing all of the code inside each `waitFor` block to essentially be ignored.
Other problems fixed:
* Rendering a router inside a router was causing most of the render() calls to fail (our custom `render()` already provides a router so there's no need to provide one in the test case)
* Use of `testid` instead of queries based on what users see
* Tests could match on content in the body when trying to make assertions about the header
* Mock imported via `index.js` was causing `jest-haste-map` to print warnings about duplicate mock names (this is still happening for other mocks)
* Passing `courses: null` instead of `courses: []` was causing a broken render on two of the tests.
I also made other cleanups to follow best practices.
bd18e874b5