Prioritizing due to the 300+ failures since April in issue 589753.

Note ihf's comment that the minidump_stackwalk error is also printed in a successful run: https://bugs.chromium.org/p/chromium/issues/detail?id=589753#c343

I'm looking at the latest error[1].  The autotest client refers the executable by /usr/bin/minidump_stackwalk, and the server refers by simply minidump_stackwalk.  I wonder if /usr/bin is not in PATH, or the executable is not installed to /usr/bin.  But how can I check?

[1] https://pantheon.corp.google.com/storage/browser/chromeos-autotest-results/72546702-chromeos-test/chromeos4-row2-rack10-host9/debug/

Failures in minidump_stackwalk have been going on since roughly
the dawn of time (or at least, since the dinosaurs...)  Generally,
this error is a nuisance.  It's never been known to be the cause
of an actual failure.

How do we know that that particular error is causing failures?
Do we actually know that?

I looked at a test, and the result is passed.  Since the test actually generate a crash on the client side, and probably not intentionally, the dump is copied back to the server, I start to wonder if the issue is automatically captured just because of that.

More positive signal to my theory.  The only crash dump that the server tried to call with minidump_stackwalk is crasher_nobreakpad.20160809.063718.24009.*.  Looked at the client side[2], logging_UserCrash generated many crashes as expected.  But from the log, the client always `rm -rf /var/spool/crash` before running a test, but not after.  That leaves the last crash dump in the directory, and rsync fetched them back to the server.  The "auto bug opener" probably then filed those issues since the server says the client has crash dump.

I'll try to make the client clean up more clearly.

CL is here: https://chromium-review.googlesource.com/#/c/367540/

Who would be able to help review the change?  I isolate the change to logging_UserCrash just to avoid touch the other logging_*Crash tests.  Otherwise, I could rename cleanup to after_run_once in the parent test class, crash_test.  But I don't know if it's ok.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/third_party/autotest/+/dcaa00f750ece133b603a6606848ff54729e8189

commit dcaa00f750ece133b603a6606848ff54729e8189
Author: Victor Hsieh <victorhsieh@chromium.org>
Date: Tue Aug 09 23:06:22 2016

Stop collecting intentional crash by logging_*

TEST=test_that logging_UserCrash  # No longer see INFO New Crash Dump
BUG= chromium:619171 

Change-Id: I8758a0fc32c277d6255da7a3b5df5406e8d1ffbe
Reviewed-on: https://chromium-review.googlesource.com/367540
Commit-Ready: Victor Hsieh <victorhsieh@chromium.org>
Tested-by: Victor Hsieh <victorhsieh@chromium.org>
Reviewed-by: Achuith Bhandarkar <achuith@chromium.org>
Reviewed-by: Simran Basi <sbasi@chromium.org>

[modify] https://crrev.com/dcaa00f750ece133b603a6606848ff54729e8189/client/cros/crash_test.py

Closing. Please reopen it if its not fixed. Thanks!