Issue: NAT-T interaction (#3)

This is about the interaction with NAT traversal:
does the MOBIKE protocol work one one party goes behind NAT,
comes from behind NAT, or goes from one NAT to another?
There was a lot of discussion about this at Seoul, so it
is clearly an issue of concern to many people. Is limiting
our protocol to only working if there is no NAT might be
too restrictive because most users don't know when they
are behind a NAT, or are about to move to behind a NAT?

Or should MOBIKE and NAT-T be seen as alternative
solutions; if you enable NAT-T then you use just
that and live with its limitations & if you use
MOBIKE then you don't have a NAT?

In Seoul it seemed that the WG wanted MOBIKE and
NAT-T to work together. If so, would testing for
NATs occur after each address change?

[Reminder: our charter says we shall not replace
or modify NAT traversal, and that we should not
integrate the protocol tightly with NAT traversal.
However, we need to document how (and when) they
work together.]