Upgrade rust toolchain to nightly-2023-06-20

[celinval]

Description of changes:

Change Kani's compiler and tests to adapt to changes done to the toolchain. The biggest changes were:

1. Implement overflow checks as part of rvalue.rs instead of intrinsics due to lowering of intrinsics to BinOp.
2. Add a unsupported check for handling C string literals.

Here is a list of changes that I found that impacted this upgrade:

• Implement RFC 3348, c"foo" literals rust-lang/rust#108801
• Lower unchecked_div/_rem to MIR's BinOp::Div/Rem rust-lang/rust#112168
• Promote unchecked integer math to MIR BinOps rust-lang/rust#112238
• Switch the BB CFG cache from postorder to RPO rust-lang/rust#112638
• Replace tcx.mk_trait_ref with TraitRef::new rust-lang/rust#110806
• Make EarlyBinder's inner value private rust-lang/rust#112006
• Use translatable diagnostics in rustc_const_eval rust-lang/rust#111677
• Write to stdout if - is given as output file rust-lang/rust#111626
• Use 128 bits for TypeId hash rust-lang/rust#109953
• EarlyBinder::new -> EarlyBinder::bind rust-lang/rust#112060
• Replace tcx.mk_re_* with Region::new_* rust-lang/rust#112075
• Move rustc_middle/src/ty/query.rs to rustc_middle/src/query/plumbing.rs rust-lang/rust#111625
• Box AssertKind rust-lang/rust#111082
• Avoid &format("...") calls in error message code. rust-lang/rust#111633
• Removed use of iteration through a HashMap/HashSet in rustc_incremental and replaced with IndexMap/IndexSet rust-lang/rust#110040

Resolved issues:

Resolves #2544

Related RFC:

Optional #ISSUE-NUMBER.

Call-outs:

This is still a draft PR. Before publishing it, I still need to:
1. I'll pull a bunch of the tests changes to a separate PR before publishing it.
2. This is still BLOCKED by a performance degradation on one of our tests. I created #2576 to track a performance issue caused by this upgrade. I modified the regression test to go around the performance issue and created a new test that still captures the issue and is part of the performance CI job instead.

Note: I don't think there is anything complicated to actually implement C string literal support, but decided to keep this simple for now. We would also likely want to add more thorough tests too.

Testing:

• How is this change tested?

• Is this a refactor change?

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]

For the itoa test, I narrowed it down the performance issue to a performance regression when formatting an i8. Updating the test to a u8 makes it much faster. I am adding a new performance test and modifying the itoa test for now. I created #2576 to capture the performance regression.

[adpaco]

Thanks @celinval for this huge effort!

The situation with the duplicate checks is not ideal (will be confusing for users and may be downgrading performance), but we can improve it later.

[celinval]

Thanks @celinval for this huge effort!

The situation with the duplicate checks is not ideal (will be confusing for users and may be downgrading performance), but we can improve it later.

Yeah, sorry about that. The problem already existed, and it just got even worse. Fixing it will take a bit of time. Maybe I'll at least disable the redundant Cbmc checks for now, since that's an easy fix.