Alignment: move from ptr to mem and rename as_nonzero to as_nonzero_usize

[GrigorenkoPV]

• tracking issue: Tracking Issue for ptr::Alignment type #102070
• split off from Partially stabilize ptr_alignment_type as alignment_type #153261

[rustbot]

r? @Mark-Simulacrum

rustbot has assigned @Mark-Simulacrum.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: @scottmcm, libs
• @scottmcm, libs expanded to 8 candidates
• Random selection from Mark-Simulacrum, joboet, scottmcm

[bjoernager]

Should the identifier not be as_non_zero_usize? The current convention with NonNull, at least, is to use non_null as the snake_case form.

[clarfonthey]

That's an interesting observation, and after looking into precedent… it could be either way. Searching for "non" in the standard library docs: https://doc.rust-lang.org/nightly/std/?search=non

Excluding false positives, we have the following which are stable:

• std::net::(various)::set_nonblocking
• (various)::{copy, swap}{, _from, _to}nonoverlapping
• std::fmt::(various)::finish_non_exhaustive

And the following unstable:

• std::mem::Alignment::as_nonzero (this one)
• (various)::{as, from, into}_non_null (tracked in Tracking Issue for box_vec_non_null #130364)
• std::hint::prefetch_{read, write}_non_temporal (tracked in Tracking Issue for hint_prefetch #146941)
• std::process::ExitStatusError::code_nonzero (tracked in Tracking Issue for ExitStatusError #84908)
• std::ptr::NonNull::<[T]>::as_non_null_ptr (tracked in Tracking Issue for raw slice getters (slice_ptr_get) #74265)
• std::intrinsics::ct{t,l}z_nonzero (perma-unstable)

So, really, the answer is that there's not really been a consistent choice, except that non_exhaustive is stable and so is nonblocking and nonoverlapping. The rest are just unstable methods.

In general… we do tend to prefer non_zero over nonzero, but we do have nonblocking and nonoverlapping that probably could be deprecated and renamed, if someone cared enough.