MaybeDangling: ensure references fit inside the address space

[RalfJung]

In the RFC we left open the question of the exact validity invariant for references inside MaybeDangling. This PR implements the strictest invariant I can think of: we already require references to be aligned, now we also require "addr + size" to be computable without overflow. This ensures that whatever niches we add to references in the future, MaybeDangling preserves those niches.

Cc @rust-lang/opsem @WaffleLapkin

Tracking issue: #118166
r? @oli-obk

• I did not use an LLM to create a change in this PR.
• I used an LLM to create a change in this PR, and I have explained below how it was used.

[rustbot]

Some changes occurred to the CTFE / Miri interpreter

cc @rust-lang/miri

miri is developed in its own repository. If the Miri part of this change can be broken out, consider making this change to rust-lang/miri instead. However, if Miri needs adjusting for rustc changes, just ignore this message.

cc @rust-lang/miri

Some changes occurred to the CTFE machinery

cc @oli-obk, @lcnr

[CAD97]

This is the obvious validity requirement, making the relaxation for MaybeDangling simply what's required for the nominal effect. LGTM

r+ (but I don't have r+ perms)

View changes since this review