Subtle execution timing issues as Javascript is added, moved, and removed from the DOM

[pshaughn]

WPT tests html/semantics/scripting-1/the-script-element/execution-timing/083, 128, 147, and 148 fail. They also all fail in Safari, and 083 fails in everything, so it could be a problem with text expectations or specs and not a problem with Servo.

• 083: updating a frame's location to about:blank is not killing an event listener that lived in the frame, but all the other browsers are failing too so maybe it's not really supposed to
• 128: Only Firefox is passing this test. The fact that adding Javascript text to an already-executed <script> executes that text at all already seems suspicious to me, and I am not surprised that browsers disagree about the intended execution order when someone does this thing.
• 147: Safari also fails this one, but Chrome/Edge/Firefox all pass. It's moving scripts around in the DOM and then expecting them to fire in the original order, not the changed order. (I had the expected and actual cases mixed up when I described it earlier, and as it turns out it's not exemplary of a larger case at all like I had thought it was.)
• 148: Only Firefox is passing this test. It looks like it's calling removeChild at a script and expecting the removed script to execute anyway.