Add simple regression test to ensure that setting sequence.editor will
not cause a hang, and add a timeout to interactive mode tests to ensure
they are terminated with an error should they fail to complete
reasonably quickly (suggesting a hang).
Early release 0.9.0 had an issue where setting sequence.editor would
cause git-upstream to hang, due to git-rebase attempting to launch the
editor specified, while the console was being redirected by python's
subprocess.
At some point a behaviour change meant the config setting
sequence.editor was no longer used by the python subprocess calling
git-rebase. Instead the tool would launch the user specified editor
directly (checking the git config settings for sequence.editor, then
core.editor, and falling back to the environment variable EDITOR if
neither were set), and would instead ensure that when finally calling
git-rebase it would only use the special editor provided by
git-upstream.
However it's still important to check that setting sequence.editor
does not cause any future problems such the code be refactored again.
Change-Id: I303dced0d1a52791d18b39e8e6d342374e3d2eeb
Related-Bug: #1394553
dfa212ffd0