Added support for improvement_direction to relevant values, which is propogated to chartjson.

tools/perf/measurements/page_cycler.py

 # Copyright 2012 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 """The page cycler measurement.
 
 This measurement registers a window load handler in which is forces a layout and
 then records the value of performance.now(). This call to now() measures the
 time from navigationStart (immediately after the previous page's beforeunload
 event) until after the layout in the page's load event. In addition, two garbage
 collections are performed in between the page loads (in the beforeunload event).
 This extra garbage collection time is not included in the measurement times.
 
 Finally, various memory and IO statistics are gathered at the very end of
 cycling all pages.
 """
 
 import collections
 import os
 
 from metrics import cpu
 from metrics import memory
 from metrics import power
 from metrics import speedindex
 from metrics import v8_object_stats
 from telemetry.core import util
 from telemetry.page import page_test
+from telemetry.value import improvement_direction
 from telemetry.value import scalar
 
 
 class PageCycler(page_test.PageTest):
   def __init__(self, page_repeat, pageset_repeat, cold_load_percent=50,
                record_v8_object_stats=False, report_speed_index=False,
                clear_cache_before_each_run=False):
     super(PageCycler, self).__init__(
         clear_cache_before_each_run=clear_cache_before_each_run)
 
@@ skipping 80 matching lines @@
                          'warm_')
 
     results.AddValue(scalar.ScalarValue(
         results.current_page, '%stimes.page_load_time' % chart_name_prefix,
         'ms', tab.EvaluateJavaScript('__pc_load_time'),
         description='Average page load time. Measured from '
                     'performance.timing.navigationStart until the completion '
                     'time of a layout after the window.load event. Cold times '
                     'are the times when the page is loaded cold, i.e. without '
                     'loading it before, and warm times are times when the '
-                    'page is loaded after being loaded previously.'))
+                    'page is loaded after being loaded previously.',
+        improvement_direction=improvement_direction.DOWN))
 
     self._has_loaded_page[page.url] += 1
 
     self._power_metric.Stop(page, tab)
     self._memory_metric.Stop(page, tab)
     self._memory_metric.AddResults(tab, results)
     self._power_metric.AddResults(tab, results)
 
     self._cpu_metric.Stop(page, tab)
     self._cpu_metric.AddResults(tab, results)
@@ skipping 15 matching lines @@
 
   def ShouldRunCold(self, url):
     # We do the warm runs first for two reasons.  The first is so we can
     # preserve any initial profile cache for as long as possible.
     # The second is that, if we did cold runs first, we'd have a transition
     # page set during which we wanted the run for each URL to both
     # contribute to the cold data and warm the catch for the following
     # warm run, and clearing the cache before the load of the following
     # URL would eliminate the intended warmup for the previous URL.
     return (self._has_loaded_page[url] >= self._cold_run_start_index)