Ramkumar Ramachandra authored Dec 14, 2011 and Junio C Hamano committed Dec 15, 2011

The "cherry-pick persists opts correctly" test in t3510
(cherry-pick-sequence) can cause some confusion, because the command
actually has two points of failure:

1. "-m 1" is specified on the command-line despite the base commit
   "initial" not being a merge-commit.
2. The revision range indicates that there will be a conflict that
   needs to be resolved.

Although the former error is trapped, and cherry-pick die()s with the
exit status 128, the reader may be distracted by the latter.  Fix this
by changing the revision range to something that wouldn't cause a
conflict.  Additionally, explicitly check the exit code in
"cherry-pick a non-merge with -m should fail" in t3502
(cherry-pick-merge) to reassure the reader that this failure has
nothing to do with the sequencer itself.

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>