Restarting the discussion on HTTP/2 stream priorities

As discussed in Seattle's interim meeting, there are some existing issues
with the current stream prioritization. There are a few things worth
discussing:

* HTTP/2's current stream priorities lacks a mechanism to correctly
multiplex stream priorities from different sources onto the same
connection. That's due to a lack of a priority context for streams. For
more on this, please refer to the previous discussion on this issue at
http://lists.w3.org/Archives/Public/ietf-http-wg/2013JanMar/0554.html. Mark
also provides a good concrete example of the issue at
http://lists.w3.org/Archives/Public/ietf-http-wg/2013JanMar/0560.html.
* Whether or not it's desirable to provide richer prioritization semantics.
For example, better mechanisms for reprioritization. Using integer stream
priorities has different tradeoffs here versus specifying stream priorities
as dependencies (thus allowing creating an ordered list/tree).

To rehash old discussions, I believe most folks found the former point
definitely worth fixing, although I don't believe we've reached any
consensus on the exact mechanism for that. I think that the second point is
far more controversial and requires more discussion.

An old, but very pertinent proposal we wrote up for SPDY/4 can be found
here:
https://docs.google.com/document/d/1pNj2op5Y4r1AdnsG8bapS79b11iWDCStjCNHo3AWD0g/edit.
It's useful to provide a concrete straw man proposal for discussion,
notably replacing the stream integer priority with a stream dependency.
Note that the proposed REPRI frame already exists in HTTP/2 as the PRIORITY
frame (http://http2.github.io/http2-spec/#PRIORITY).

I don't expect us to reach the point where a discussion in Vancouver would
be fruitful, but I'd be very happy if we can progress this discussion to a
point where we can come to a consensus by the next interim meeting.