HSTS Priming, continued.

The discussion at TPAC more or less convinced me that Richard's HSTS
Priming proposal is a good one that we should explore. To that end, I put
together a little bit more detail about what it would mean for Fetch and
etc. so that we have more details to argue about.

I'd appreciate it if you'd take a quick look at
https://mikewest.github.io/hsts-priming/ and give some feedback about those
details (Note: I think the normative changes should end up in Fetch and
HSTS and MIX, but it's easier to discuss a single document than a set of
diffs).

CCing some folks who I think might be particularly interested.

-mike