Apply changes suggested by Barry Leiba.

[huitema]

Close #13

[huitema]

This PR takes into account the changes suggested by Barry. Most of the changes are simple, but we need to check two points:

1. I placed MDNS and DNSSD as normative references. Are there other references that should be classified as such?

2. Barry observed that using MDNS as the reference for sleep proxies was very loose, and he is right. The problem is that there are very few good publications describing sleep proxies. When I checked that, I mostly got references to specs by Intel or Apple that were shared with implementers under NDA. I found an old web page by @StuartCheshire describing the sleep proxy on the MAC and I used that, but I would prefer an open spec or an academic reference.

[huitema]

Also addressed the comments by Roman Danyliw.

Close #15

[huitema]

Close #16

[huitema]

Close #17

[huitema]

At this point, the PR is ready and addresses pretty much all the comments. Waiting for reviews before submitting before the IETF cut-off time.

[huitema]

Checking in, as draft 04 was submitted.