Enable hex printer in cargo-miri-test

[RalfJung]

Once rust-lang/rust#52744 lands

We probably should also add memrchr to the main test suite, then.

[RalfJung]

It occurred to me that my patch would not be sufficient, and indeed...

error[E0080]: constant evaluation error: tried to access memory with alignment 1, but alignment 8 is required
   --> /home/r/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/src/libcore/slice/memchr.rs:119:21
    |
119 |             let u = *(ptr.offset(offset as isize - 2 * usize_bytes as isize) as *const usize);
    |                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ tried to access memory with alignment 1, but alignment 8 is required
    |
note: inside call to `core::slice::memchr::memrchr`
   --> tests/run-pass-fullmir/memchr.rs:69:39
    |
69  |         assert_eq!(Some(pos - start), memrchr(needle, &data[start..]));
    |                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
note: inside call to `each_alignment_reversed`
   --> tests/run-pass-fullmir/memchr.rs:89:5
    |
89  |     each_alignment_reversed();
    |     ^^^^^^^^^^^^^^^^^^^^^^^^^

error: aborting due to previous error

miri puts the entire slice into the first argument, leading memrchr to think that most of it is actually aligned.

I'll have to tweak the implementation some more.

[RalfJung]

Opened rust-lang/rust#52854

[RalfJung]

Fixed by #423