Thank you for the detailed edits. These changes are worth incorporating. Perhaps even if CSRC, SSRC etc. are not always needed ‘at the root of the definitions’, it makes the specification text cleaner.
You are right. We should say that the extension boxes (pasp, clap etc.) defined in part 12 should FOLLOW any boxes required by derived specs, not precede them. We should also suggest that readers cope with any box order.