Removing architectures for 16.0

[markjdb]

We need to publicize any removals well in advance, ideally in the 15.0 release notes.

Potentially on the chopping block, based on the number of users we think there are, are armv7 and powerpc64.

John lists some maintenance costs:

• pmap (pv entry handling is unique in some powerpc pmaps)
• rtld
• newbus/driver resource handling code is hard to test when doing tree sweeps

powerpc64 does still have some maintainership.

[markjdb]

@bsdjhb and @bsdimp seem to be wiling to take this on...

[markjdb]

20251128: armv7 and powerpc64 discussions have started; it seems that there's a fair bit of interest in keeping powerpc64le, and there's some pushback against removing armv7 in 16.0 as well.

[bsdimp]

Yes. We've all but committed to powerpc64le. So that's a done deal for now.
It's hard to decommit to armv7 at this time. @bsdjhb and I seem to be searching for the right words to say "armv7 may or may not be removed in 16. It's on a knife's edge, since we need more people to devote more resources to keep the port viable. If that happens, then it will be retained. If those resources aren't applied, it will be removed." with the right amount of nuance to scare people enough to put forward the effort, while not scaring them too much so they go away faster than they would have otherwise.

[markjdb]

We might wait until 15.1 to make a decision on armv7. We haven't seen a new maintainers step up the way we have for powerpc64, but there is some resistance to removing support. If no one steps up in the next few months, the case for removing armv7 will be much stronger.