Machine-readable copyright module, take 2

[John Wright]

I'm re-sending the whole patch stack, since there were some (minor)
changes in the earlier patches as I wrote the rest of them.

This contains the bulk of functionality.  What remains (and I haven't
started any work on yet), is
  - Parsing the License synopsis as a list of alternatives, where
    applicable.  I haven't thought very hard about the API to expose
    yet, but it seems similar in some ways to the package relations
    stuff.
  - Look-up of a license text once a choice has been made.  E.g. I have
    a FilesParagraph, which has a License field with only a list of
    alternatives (which are then spelled out later in License
    paragarphs).  I might want to know the full text of a particular
    choice.  Not too hard in principle, but I want to get the API right.
  - Look-up of license texts from common-licenses might be a nice thing
    to do as well.  Meanings of short-form license names are technically
    tied to the value of the Format header field, so this could be a bit
    tricky (more than 1/3 of the machine-readable copyright files in the
    archive use a format other than the current one).

In any case, I actually think this is the functionality that most users
who would want a copyright parser actually need.

If you prefer a graphical view with some (not great, but existing)
support for commenting on patches, I've created

  https://gitorious.org/python-debian/python-debian/merge_requests/1