This does several things:
- Split the test up, so one assert doesn't fail the entire test suite
- Check preload, playbackRate and defaultPlaybackRate attributes after unsetting srcObject
- Check setting currentTime before loading (default playback start position),
  and after loading (official playback position) separately
- Check that duration is changed to a real number on ending
- Check that HAVE_ENOUGH_DATA is reached by only assigning srcObject
- And other minor things and formatting

Differential Revision: https://phabricator.services.mozilla.com/D33647

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1554699
gecko-commit: a6187e39bf7c0ba5d9ba36d570a1b6d8fc4f7ecd
gecko-integration-branch: autoland
gecko-reviewers: jib

…ing a MediaStream.

Differential Revision: https://phabricator.services.mozilla.com/D33655

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1554699
gecko-commit: 7727cef089cd73f8dc0f76afd2b8309603470709
gecko-integration-branch: autoland
gecko-reviewers: jib

…ved.

Since the media element is playing directly after play(), not waiting for the
readyState to advance before play()ing could leave the play promise from play()
unhandled until cleanup, when it would be rejected by the load algorithm.

This would happen if a timeupdate event is raised before the first frame comes
in to update the readyState, which resolves the play promise.

Waiting for the readyState to advance before play()ing seems like the cleanest
fix wrt testing currentTime after the first timeupdate event.

Depends on D33655

Differential Revision: https://phabricator.services.mozilla.com/D34686

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1554699
gecko-commit: 89350b209bcecaa0f0ec46769647cee3f0d8084b
gecko-integration-branch: autoland
gecko-reviewers: jib