The tests that return files, symlinks, directories, and empty dirs were
all only testing that the type of the return data was the same as what
was expected. By not testing the content, we overlooked a corner case in
which backends passed into the fileserver as a Python list would not be
handled properly. This has since been fixed in PR #36244, but these
tests will help keep this sort of issue from regressing.
The return data from run_run_plus previously returned the runner return
in the ``fun`` key. This has since been corrected. This commit replaces
these references to the ``fun`` key with the corrected ``return`` key.
The change in runner output behavior caused the output of test_envs to
change, causing some changes to need to be made in da163471 to get them
to pass. Now that runner output is fixed, the test was returning the
output it should have been in the first place. This commit changes that
test back so that it now expects a list as output.
Additionally, I've taken the opportunity to add additional test calls to
test the "backend" argument for those functions which now have it.
