Add unchecked/SIMD bitshift checks and disable CBMC flag

[reisnera]

Description of changes:

The CBMB flag --undefined-shift-check is too strict. In particular, it flags shifting a negative value as being incorrect, which is not the case for Rust where shifts are valid unless the shift distance is negative or larger than the bit width of the value being shifted.

This PR disables the CBMC flag and adds checks to Kani both for the unchecked shift BinOps as well as SIMD shift intrinsics.

Resolved issues:

Resolves #2374 - shifting a negative value now works.
Resolves #1963 - SIMD shifts now have the same checks as non-SIMD-shifts, namely that the shift distance is not negative or larger than the bit width of the value being shifted.

Related RFC:

n/a

Call-outs:

The changes in intrinsic.rs could probably be more DRY but I'm not sure it's worth it.

Testing:

• How is this change tested?
  I believe existing tests cover this PR.

• Is this a refactor change?
  No

Checklist

• Each commit message has a non-empty body, explaining why the change was made
• Methods or procedures are documented
• Regression or unit tests are included, or existing tests cover the modified code
• My PR is restricted to a single feature or bugfix

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 and MIT licenses.

[celinval]

It looks pretty good. Can you please add a test to ensure that CBMC does the expected thing for shift with negative values?

Something like:

#[kani::proof]
#[kani::unwind(32)]
fn check_shl() {
    let val : i32 = kani::any();
    let dist = kani::any_where(|d: &u8| *d < 32); 
    assert_eq!(val << dist, val.wrapping_mul(2_i32.wrapping_pow(dist.into())));
}

you might also want to add a modified version of the test you created in the issue description.

[reisnera]

Will work on adding the tests when I can! Also, my clock somehow got out of sync so I apologize about any incorrect date/timestamps on these commits.

[celinval]

Thanks

[adpaco-aws]

Thank you, @reisnera !

my clock somehow got out of sync so I apologize about any incorrect date/timestamps on these commits.

No worries, we use the "squash and merge" option for PRs.