Treat "skipped" unit tests as less interesting than "passed"
Ref T10457. Skipped tests are almost always well-behaved (e.g., testWindows(), but the test is running on Linux) and not interesting, and we do not expect well-written, solid systems to necessarily have 0 skips.
Although skips could indicate that you have missing dependencies on a build server, and thus be a bit interesting, I think they almost always indicate that a particular test is not expected to run in the current environment.
If we wanted to tackle this problem in granular detail, we could eventually add a "Missing" status or similar which would serve as "a skip you could reasonably fix in this environment", but I don't think that's too interesting.
Here's an example of a build result with skips: B10875
I think this is clearer as "Passed", as this is the expected production state of the build.
Locally, looked at some builds.
Reviewed By: chad
Maniphest Tasks: T10457
Differential Revision: https://secure.phabricator.com/D15369