Audit for arith_offset intrinsic

[adpaco-aws]

Description of changes:

This PR adds support for the arith_offset intrinsic (used by wrapping_offset) by using the already-implemented codegen_offset function.

The documentation for wrapping_offset defines it as a more permissive version of offset which allows the computed offset to go out-of-bounds. Currently, we don't check in codegen_offset that the computed offset is within bounds to avoid spurious failures, so its definition should work for both.

Resolved issues:

Part of #1163

Call-outs:

Testing:

• How is this change tested? Adds 6 tests.

• 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.

[tedinski]

There's a note in your PR description that isn't reflected in the comments of codegen_offset (that these two instrinsics should have different behavior?)

Should there be a tracking issue for this? If so, can you create that and link to it from that function's comments too?

Approving because with that minor change to comments, LGTM

[adpaco-aws]

Thanks @tedinski , you're right that we should be tracking this.

I created #1233 and extended the comment to both link that issue and clarify that it's only concerning offset. If we had to add that check at some point, it would be enough to add a flag to check for this property in codegen_offset.