force `subprocess.run` to check return code is 0

In general, this should help to capture unintentional errors.

Where the change has created failures, they have been fixed. (Mostly because the git repository that was looked for was missing.)

tests.py

@@ -31,7 +31,7 @@ def test_prepare_environment(tmp_path):
     # prepare_env is run when cmd is not specified, we can test cmd='prepare'
     # and cmd=None to check both cases
 
-    script_dir = tmp_path / "test_prepare_env"
+    script_dir = tmp_path / "mpsd_opt" / "linux_debian_11"
     spack_environments = "spack-environments"
     mpsd_release_to_test = "dev-23a"
     release_base_dir = script_dir / mpsd_release_to_test