New Arch: armv7

[Warner Losh]

As discussed here in June, we're creating a new arch.

I've landed on the armv7 name (as opposed to earmv7hf or armv7l) for the
following reasons:

1. armv7 isn't a huge amount of work to implement in our tree, while the
other two names require substantially more complicated .if statements in
makefiles.
2. Neither of the other names actually buy us a large amount in the ports.
Autoconf is split and many other ports do different things to get the
system they are running on and wouldn't benefit from the other names.
There's a small number of ports that might be better, but not enough to
justify the extra work in base.
3. It's more like the names we've been using, and won't cause confusion
with our user base. And it will be easy enough for outsiders to see what we
support from it without needing the decoder ring for NetBSD, or dealing
with 'what's the trailing l mean, why doesn't armv6 have it, or arm and
what about armeb' all the time.

btw, if we ever do a big endian port, armv7eb will be the name, per project
tradition.

So, I'll be finalizing FCP-0100 with this data (I've sent the pull request)
and will get an implementation together (I have the start of one I was able
to knock out in about an hour, after spending twice that on the other two
choices w/o reaching completion) and if there's issues that arise as we
move forward, cope with them on the way to asking core to vote to bless
this is the consensus of the relevant parts of the community.

I hope to have it committed by the end of the month, and hope that the FCP
process won't unduly delay things.

Warner