We have a set up where potentially hundreds of tests run when doing arc diff. When the unit tests fail, people tend to put in an excuse because it's easier to view the unit test results in the web UI (viewing only the failed tests), than trying to scroll up through hundreds of lines of output to try and find the tests that failed.

Instead, Arcanist should reproduce a copy of the summary of all failed tests so that users can easily see what tests failed and what the failure was.