Finish macOS arm64 SHA assembly support

[busterb]

This is just PR #1226 rebased on master and with a few additional fixes applied on top.

[botovq]

I think you'll need an ok from @4a6f656c for this.

Unless I'm missing something, the crypto_assembly.h part clobbers the upstream one. That should probably be a patch instead (too close to OpenBSD full lock).

Edit: I'd probably prefer landing this early in the next cycle, but maybe it's trivial enough. I don't know.

[busterb]

You're right, there's logic now in update.sh that prevents upstream's minimal crypto_assembly.h from clobbering this one. It could be a patch too, I was just copying the pattern that was in @4a6f656c 's original draft PR. If they don't have time to review, I can just drop this for now.

[busterb]

Sounds like this isn't going to happen. I'll just close it.

[botovq]

It didn't sound like anything...

I really don't like seeing non-trivial effort go to waste like this. It's very frustrating. If this has no chance of landing, it would be nice if @4a6f656c could say so quickly instead of letting things rot for such a long time. On the other hand, if it does have a chance of landing, it would also be nice to have some sort of feedback in that direction.

[4a6f656c]

Apologies for not seeing this earlier. The majority of the changes are now upstreamed (which was always the plan), so the only part needed should be the build glue.