Rough draft of WOFF2 spec

Hi all,

   Apologies for taking so long, but finally I have a draft of the WOFF2
wire format that I think is useful to share.

   I'll be checking this in to W3C's version control system soon, so we can
apply proper processes for maintaining it. In the meantime, I'm very happy
to get comments and feedback. In particular, right now I'd particularly
like to hear people's take on which parts (if any) are reasonably solid,
and which need more work, and in the latter case what kinds of improvements
are most important.

   The triplet coding in particular is not filled in, and the draft
contains a reference to the MTX submission. However, that is an
informative, not a normative reference, and ultimately the WOFF 2 spec
should contain a full description and not rely on the MTX submission. Vlad
has generously volunteered to work on this part. I consider it one of the
least "tricky" or controversial aspects of the spec, because the triplet
encoding itself is identical to MTX and there are multiple independent
implementations of that.

   Also speaking of normative referenecs, right now, the [Brotli] reference
links to the draft spec in the open source release of Brotli. I think this
is useful for a rough draft, but obviously this will need to be replaced
with the real reference as soon as that spec is properly published. I hope
to have an update on the status of the Brotli spec by tomorrow's call.

   Thanks to Vlad, David, and Chris Lilley for help and encouragement
getting this draft written, and of course everybody who's been
participating.

   I look forward to discussing this draft on the call tomorrow.

Raph