[benchmark] loading: Report the UMA metrics.

Context: https://docs.google.com/document/d/1fQM5XxoPxGGQfjpQhKe6No9eVoQic2-1MGzgSgKhOOM/edit?hl=en#
Summary of pinpoint runs: https://docs.google.com/spreadsheets/d/1DXm_xeZu8xEP5dZEszqAcIyjPJsd7p3crb-cS9Z7tYU/edit#gid=0

Interesting observation: the UMA metric for FirstContentfulPaint match
very closely with the tracing-based metric for _cold loads. However, for
_warm/_hot loads, the UMA metric is higher than the tracing-based
metrics in a majority of the stories (although there are a couple of
exceptions).

Another observation: it looks like the LargestContentfulPaint UMA metric
is not reported at all for _cold loads. It is unclear whether this is a
bug in reporting, or whether the test ends too quickly for the _cold
loads.

BUG=1049117

Change-Id: Ic213eacc537a6430b8cbe3f6d6ef8491370c69d7
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2037962
Commit-Queue: Sadrul Chowdhury <sadrul@chromium.org>
Reviewed-by: Annie Sullivan <sullivan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#738614}

tools/perf/benchmarks/loading.py

@@ -25,7 +25,9 @@ class _LoadingBase(perf_benchmark.PerfBenchmark):
     # memory used throughout the page loading tests. Include "umaMetric" as a
     # timeline so that we can parse this UMA Histogram.
     tbm_options.config.chrome_trace_config.EnableUMAHistograms(
-      'Memory.GPU.PeakMemoryUsage.PageLoad')
+      'Memory.GPU.PeakMemoryUsage.PageLoad',
+      'PageLoad.PaintTiming.NavigationToLargestContentfulPaint',
+      'PageLoad.PaintTiming.NavigationToFirstContentfulPaint')
 
     # Add "umaMetric" to the timeline based metrics. This does not override
     # those added in loading_metrics_category.AugmentOptionsForLoadingMetrics.