My thinking is that this may reduce the cognitive overhead for
developers writing new test suites.
This also allows us to perform different setup steps (like ensuring the
editor is visible when authenticated).
This uses the current dev build by default but can be overidden with
CODE_SERVER_TEST_ENTRY (for example to test a release or some other
version).
Each instance has a separate state directory. This should make
parallelization work.
This also means you are no longer required to specify the password and
address yourself (or the extension directory once we add a test
extension). `yarn test:e2e` should just work as-is.
Lastly, it means the tests are no longer subject to yarn watch randomly
restarting.
It errors that jest is not defined so put it behind a function instead
of immediately creating the mock (this is probably a better pattern
anyway).
The constant tests had to be reworked a little. Since the logger mock is
hoisted it runs before createLoggerMock is imported. I moved it into a
beforeAll which means the require call also needed to be moved
there (since we need to mock the logger before requiring the constants
or it'll pull the non-mocked logger).
This means getPackageJson needs to be a let and assigned afterward. To
avoid having to define a type for getPackageJson I just added a let var
set to the type of the imported constants file and modified the other
areas to use the same paradigm.
I also replaced some hardcoded strings with the mocked package.json
object.
This is to match the other tests that create temp directories. It also
lets you clean up test temp directories all at once separately from
other non-test temporary directories.
