I saw in the past that if we started too many DRTs at once while running parallel tests, the first few tests would time out (probably caused by DRTs thrashing on something while initializing). This led to me putting a time.sleep() into new-run-webkit-tests in between starting workers. I'm filing this bug to track down what's going on, and filing another bug to work around this in a way that won't slow down the unit tests in the meantime.
Filed bug 79148 to not sleep when running the TestPort, so that unit tests are faster ...