This test has been failing a lot randomly (but not every time) on Leopard and SnowLeopard Release buildbots since r53514-r53517. However, it failed occasionally (though less often) even before that, for instance see <http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r53494%20(9561)/results.html>. Thus, it appears there is something wrong with either this test or the code it is testing that leads to random failures.
I'm taking a look at this to see if I can make it more stable.
Another instance: https://bugs.webkit.org/show_bug.cgi?id=39128 From both the test codes, I think they are from the same cause.
I don't see any failures in recent past. No TestExpectations exist for this test at the time.