Speaking as a WG participant:
The latest version looks good to me. A couple of new issues:
1).
>Usage: foreverypart [":name" string] block
The document is silent on whether each name string should be unique or
not. What should happen if 2 nested foreverypart loops use the same
name: the inner loop's name hides the outer name or is this a runtime error?
2). I think the MIME Loops document should discuss implications on
digital signatures. This issue came up during IESG review of several
documents recently.