add only_if_modified param to GSUtils.upload_file()

add only_if_modified param to GSUtils.upload_file()

Needed so that the file history within gs://chromium-skia-skp-summaries will indicate when results changed.

BUG=skia:1942
R=borenet@google.com

Committed: https://skia.googlesource.com/common/+/122ea7c

[epoger, 2014-07-23 13:47:27]

Eric, PTAL as of patchset 2.

I suggest that you review the two patchsets separately... I think it will be
clearer that way.

[borenet, 2014-07-23 14:55:04]

LGTM.  Thanks for dividing the rename from the behavioral change.

https://codereview.chromium.org/411723002/diff/1/py/utils/gs_utils_manualtest.py
File py/utils/gs_utils_manualtest.py (right):

https://codereview.chromium.org/411723002/diff/1/py/utils/gs_utils_manualtest...
py/utils/gs_utils_manualtest.py:67: time.sleep(1)
Is the timestamp in units of seconds?  If so, we may want to wait for 2 seconds.

[epoger, 2014-07-23 15:02:30]

https://codereview.chromium.org/411723002/diff/1/py/utils/gs_utils_manualtest.py
File py/utils/gs_utils_manualtest.py (right):

https://codereview.chromium.org/411723002/diff/1/py/utils/gs_utils_manualtest...
py/utils/gs_utils_manualtest.py:67: time.sleep(1)
On 2014/07/23 14:55:04, borenet wrote:
> Is the timestamp in units of seconds?  If so, we may want to wait for 2
seconds.

The timestamp is a string, like this:
"Wed, 23 Jul 2014 14:59:36 GMT"

Crazy, right?

I think 1 second would work consistently, as long as it really sleeps for at
least one second... but little harm in making it sleep(2) instead.  Done.

[epoger, 2014-07-23 15:02:46]

Committed patchset #3 manually as r122ea7c (presubmit successful).