Fix #250 - Implement approach for scaling up service worker termination testing

This MR introduces a new kind of tests to help us in identifying and reproducing when our existing functional tests would fail if a service worker is suspended by the browser.

The new kind of tests, which I'm calling fuzzing tests, run the same functional test suite except with the addition of suspending the service worker every time a message is sent to the background service worker.

Many tests fail when this restriction is added (I saw about 75 test failures initially), which is a good sign for how useful it will be to verify that we're making progress towards being able to confidently say that our code works even when the browser suspends the service worker. To limit the scope of this issue, most tests are currently ignored when the fuzz tests are run. This particular MR fixes the tests for uses of the filters API, excluding events.