[Telemetry] Fix handling for two-part per-page value names in Chart JSON.

[Telemetry] Fix handling for two-part per-page value names in Chart JSON.

The perf dashboard expects that for values of the form x.y that are per-page,
the computed summary x.y (i.e., the value having that name but not associated
with any particular Page object) will be displayed but the per-page x.y's will
be ignored.

Previously, the per-page x.y's were being named as x: <pagename> in the
buildbot convention; it was the job of the log processing step to make sure
that these extraneous, ambiguously-named values were ignored. The Chart JSON
output formatter, on the other hand, explicitly guards against adding values
with duplicate names, and so extra checks are necessary to throw away such
extraneous values.

This patch adds handling to the Chart JSON output formatter to address this
issue, and updates the tests to check for correctness in this scenario.

R=sullivan,qyearsley,tonyg

Committed: https://crrev.com/fd93e29f1f5d03eaf07af9b20f64a568997e874c
Cr-Commit-Position: refs/heads/master@{#294683}

[tonyg, 2014-09-10 23:31:47]

rubber stamp owners lgtm, but please wait for quinten or annie to review too.

[qyearsley, 2014-09-10 23:51:36]

I don't think I understand the handling of result names and how we want it to
be.

 - Under what circumstance is a value name of the form "x.y". Is it when there's
a "chart name" specified, like page_load_time (chart name warm_times or
cold_times)? Or when it's a per-page result?
 - Are page names included in the name property of a value?
 - What is name_suffix?

[qyearsley, 2014-09-10 23:51:57]

https://codereview.chromium.org/551063005/diff/1/tools/telemetry/telemetry/re...
File tools/telemetry/telemetry/results/chart_json_output_formatter_unittest.py
(right):

https://codereview.chromium.org/551063005/diff/1/tools/telemetry/telemetry/re...
tools/telemetry/telemetry/results/chart_json_output_formatter_unittest.py:70: d
= chart_json_output_formatter._ResultsAsChartDict( # pylint: disable=W0212
Note about pylint and testing private members:

In the past for unit tests, I've put the pylint disable comment at the top of
the unit test file with a note that the unit test is testing private members of
the module or class under test... which seems to be OK in unit tests in general.
(Not sure if anyone has any opinions/experience about when this might be bad?)

[eakuefner, 2014-09-11 21:15:52]

Quinten PTAL -- I've updated the strategy to simply overwrite an already-present
trace, which is what we discovered the graphing log processor is doing right
now.

I've not changed the unittest private interface thing because I plan to update
the test in a separate CL to construct a dummy output formatter so that we don't
have to test directly against the private interfaces. Thus I'll update the calls
one by one as I remove them.

[qyearsley, 2014-09-11 23:45:58]

Alright, LGTM now (with one nit before submitting).

https://codereview.chromium.org/551063005/diff/20001/tools/telemetry/telemetr...
File tools/telemetry/telemetry/results/chart_json_output_formatter.py (right):

https://codereview.chromium.org/551063005/diff/20001/tools/telemetry/telemetr...
tools/telemetry/telemetry/results/chart_json_output_formatter.py:48: 
Nit: only one blank line here.

[eakuefner, 2014-09-12 20:23:47]

The CQ bit was checked by eakuefner@chromium.org

[commit-bot: I haz the power, 2014-09-12 20:32:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patchset/551063005/40001

[eakuefner, 2014-09-12 20:38:45]

The CQ bit was unchecked by eakuefner@chromium.org

[eakuefner, 2014-09-12 20:42:23]

The CQ bit was checked by eakuefner@chromium.org

[commit-bot: I haz the power, 2014-09-12 20:50:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patchset/551063005/60001

[commit-bot: I haz the power, 2014-09-12 22:38:20]

Committed patchset #4 (id:60001) as 056e2b7f851195d6a4fc6b12aaa72b91739b52a0

[commit-bot: I haz the power, 2014-09-12 22:44:46]

Patchset 4 (id:??) landed as
https://crrev.com/fd93e29f1f5d03eaf07af9b20f64a568997e874c
Cr-Commit-Position: refs/heads/master@{#294683}