Auto Bisect: Added some real-world examples to confidence score unit test.
These were collected by Jeff last week: https://docs.google.com/a/google.com/document/d/1erjepFVXplUFXCybBd8aojjEDgoDGfM4V-j_GewPWOg/edit
In the examples that he collected, I think that most of the confidence score numbers were from the old confidence function. Some of the actual results still don't fully line up with intuition, for example:
[67, 68], [65, 65, 67] -> 80%
[3, 3, 3], [2, 2, 2, 3] -> 90%
In general, the t-test is always giving either 0 or at least 80 (sort of like how it is for alerts the dashboard).
BUG=
Review URL: https://codereview.chromium.org/
481063002
Cr-Commit-Position: refs/heads/master@{#291559}
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@291559 0039d316-1c4b-4281-b951-d872f2087c98