Support navigator.doNotTrack

[ghostwords]

Fixes #1835, fixes #1260.

For example, see this Medium blog page with a bunch of JavaScript DNT checks.

[ghostwords]

Using the "spam frames with messages every ten milliseconds until the frame responds or message limit is hit from chrome.webNavigation.onCommitted" approach (patch you can git apply to this branch) results in injection typically around 1 ms (but never at 0 ms) in Chrome and anywhere between 3 and 20 ms in Firefox using my navigator.doNotTrack test fixture.

Using our current "content script messages background page and acts upon response" approach, I see timings that are about twice as slow (8 ms vs. 16 ms in Firefox, for example).

Since both approaches are relatively fast while neither succeeds at intercepting the inline script (at "0 ms"), I see no reason to switch to the onCommitted approach.

[ghostwords]

Using chrome.tabs.executeScript triggered from chrome.webNavigation.onCommitted (patch) results in similar timings as the "repeatedly attempt script injection by messaging the content script upon chrome.webNavigation.onCommitted" approach above.

[bcyphers]

Looks good, I haven't been able to find any bugs. The test script gist does sometimes display 'unset (navigator.doNotTrack="unspecified")' when I open it in firefox with privacy badger enabled, but this has not caused the tests to fail for me. Maybe we should change the test page's timeout from 50 to 100 ms, just to be safe?