Update rust toolchain to 2023-04-29

[zhassan-aws]

Description of changes:

Update toolchain to 2023-04-29. Relevant PRs that required changes:

Removal of WithOptConstParam: rust-lang/rust#96840
Moving of IndexVec into its own module: rust-lang/rust#110539
Hashes use their own type instead of u64: rust-lang/rust#110083
New OffsetOf variant in rustc_middle::mir::NullOp: rust-lang/rust#106934
Lowering of offset intrinsic to BinOp::Offset: rust-lang/rust#110822

Also had to add an isize suffix to the second argument to the offset intrinsic due to rust-lang/rust#111699.

Resolved issues:

Towards #2451

Related RFC:

Optional #ISSUE-NUMBER.

Call-outs:

Testing:

• How is this change tested? Current regressions

• 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.

[zhassan-aws]

Keeping this as a draft because there are still a few failures in the expected test suite. Also, the codegen_firecracker.sh regression doesn't terminate after running for more than 10 minutes.

[zhassan-aws]

All regressions now pass except for two perf regressions that are running out of memory 😞

tests/perf/btreeset/insert_multi
tests/perf/btreeset/insert_same

[zhassan-aws]

On one of the BTreeSet performance tests (tests/perf/btreeset/insert_any), the number of nondets in CBMC's --program_only output increased significantly with the toolchain update:

$ grep -c nondet before.po 
117
$ grep -c nondet after.po 
692

The size of the program expression also increased > 5X:

$ grep "program expression" before.po 
size of program expression: 15889 steps
$ grep "program expression" after.po 
size of program expression: 87287 steps

@tautschnig is there a way to pinpoint what is causing the difference?

[zhassan-aws]

I narrowed it down to a change that happened between the 2023-04-29 and 2023-04-30 nightly toolchains: just changing the toolchain to 2023-04-29 in this PR restores performance on those btreeset tests.

[zhassan-aws]

For the sake of making progress on the toolchain updates, I've backtracked the toolchain date by 1 day to 2023-04-29 to avoid the performance regression that occurs with the 2023-04-30 toolchain.

The PR is now ready for review.

[celinval]

Thanks Zyad! I just have questions about the changes to offset implementation.