Saw issues in the CI where reset_gitlab() would fail. It would fail to
delete the group that is created when GitLab starts up. Extending the
timeout didn't fix the issue.

Changed the code so now it will check if the item is deleted and if it
isn't it will call the delete() method again to ensure that GitLab
knows it should be deleted. Since making this change I have not been
able to reproduce the failure in reset_gitlab().

Also added some logging functionality that can be seen if logging is
turned on in pytest.

I have found trying to debug issues in the functional tests can be
difficult. Especially when trying to figure out failures in the CI
running on Github.

Add logging to `tests/functional/conftest.py` to have a better
understanding of what is happening during a test run which is useful
when trying to troubleshoot issues in the CI.

If a query returns more than 10,000 records than the following values
are NOT returned:
  x.total_pages
  x.total

Modify the code to allow no value to be set for these values. If there
is not a value returned the functions will now return None.

Update unit test so no longer `xfail`

https://docs.gitlab.com/ee/user/gitlab_com/index.html#pagination-response-headers

Closes #1686

Ensure reset_gitlab() succeeds by waiting to make sure everything has
been deleted as expected. If the timeout is exceeded fail the test.

Not using `wait_for_sidekiq` as it didn't work. During testing I
didn't see any sidekiq processes as being busy even though not
everything was deleted.

chore: skip a functional test if not using >= py3.9

One of the tests requires Python 3.9 or higher to run. Mark the test
to be skipped if running Python less than 3.9.

When running with docker-compose on Ubuntu 20.04 I got the error:

  $ docker-compose up
  ERROR: The Compose file './docker-compose.yml' is invalid because:
  networks.gitlab-network value Additional properties are not allowed ('name' was unexpected)

Changing the version in the docker-compose.yml file fro '3' to '3.5'
resolved the issue.