revised requirements draft

Hi all,

Thomas and Sean asked me to make some edits to the -02 requirements
draft, drawing on what consensus I could see from the list archives
(including Michael Py's rolling summary of issues, which was very
handy).

What follows below is a candidate -03 draft. Here's a summary of the
pertinent diffs; comments most welcome. Note that this draft has not
been sent to the rfc editor yet, so if people have comments I'll
harvest the diffs paragraph by paragraph and make changes appropriately.

|3.1.7 Impact on DNS
|
|   Multi-homing solutions either MUST be compatible with the observed
|   dynamics of the current DNS system, or the solutions MUST have
|   demonstrate that the modified name resolution system required to
|   support them are readily deployable.

New section, based on suggestions from Christian Huitema and RJ
Atkinson. Brian Carpenter included a requirement that a proposed
solution should include an analysis of its impact on the DNS, which
I omitted since it seems inherent in 3.1.7 that analysis is required.

|3.1.8 Packet Filtering
|
|   Multihoming solutions MUST NOT preclude filtering packets with forged
|   or otherwise inappropriate source IP addresses at the administrative
|   boundary of the multi-homed site.
|