Collect Startup.FirstWebContents metrics on all channels.
I was running an experiment to see if reducing the bucket width of the
histograms would increase the precision of the measurements (specifically the
50th and 90th percentiles). The answer is negative. Having more buckets
resulted in a difference of less than 0.1%. This CL removes the experimental
metrics, and replaces them with Startup.FirstWebContents.MainFrameLoad and
Startup.FirstWebContents.NonEmptyPaint, using standard bucket ranges.
BUG=
Review URL: https://codereview.chromium.org/
816883003
Cr-Commit-Position: refs/heads/master@{#309964}