Update transmute_copy to ub_checks and ?Sized

[scottmcm]

The assert! in transmute_copy is from #98839. However, it was controversial at the time because it's possible to write transmute_copys that were UB by the description but didn't pass the documented conditions. It's also sub-optimal to have possible panics in unsafe code where the panic can actually make things worse by dropping things in the middle of what the programmer expected to be panic-free. Now that we have the UbChecks possibility, though, we can do this better: this PR makes it a non-unwinding panic when not meeting the documented condition when UB checks are enabled, but which very clearly emphasizes that said check isn't a documented behaviour of the function.

Also, I was staring at the signature and couldn't figure out why it needs Src: Sized. Since it always takes a reference, there seems to be no downside to accepting Src: ?Sized. That would also give it more of a reason to exist even if https://doc.rust-lang.org/nightly/std/mem/fn.transmute_prefix.html exists (as transmute_prefix(x) existing will no longer need the transmute_copy(&ManuallyDrop::new(x)) polyfill, cc rust-lang/rfcs#3844). That definitely needs an FCP, though.

A partial replacement for #154665

[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, jhpratt, nia-e

[theemathas]

Currently, this UB is detected at run time when transmute_copy is done from a small type to a large type, regardless of whether the program is compiled in debug or release mode. This PR presumably causes this UB to silently occur if the program is compiled in release mode, which seems undesirable to me.