Move to phase 3

[alexcrichton]

I wanted to open a tracking issue for remaining items before progressing to phase 3.

By the letter of the law the requirements are:

Entry requirements:

• Test suite has been updated to cover the feature in its forked repo.
• The test suite should run against some implementation, though it need not be the reference interpreter.
• Updates on the actual spec document and reference interpreter are NOT yet required (but can happen earlier).

These are satisfied with tests for wide-arithmetic and an implementation in the spec interpreter and wasmtime. The spec has also been updated at this time and is listed in the README for the relevant portions.

In that sense, this proposal could in theory be put to a vote for phase 3 at any time. Personally though I've been holding off on this to wait for toolchain support to percolate. An implementation of wide-arithmetic is available in LLVM/clang and in rustc. LLVM has not been released, however, and support will first be in LLVM 20. With LLVM's release schedule it's slated to have a 20.1.0 release in early March 2025. I believe the Rust compiler plans to update to LLVM 20.1.0 prior, though, which will be available in nightly Rust.

Once toolchain support is available for folks to more easily evaluate this proposal my plan is to put something on the CG agenda for moving this proposal to phase 3. I plan on rerunning and re-confirming all benchmarks with respect to the current proposal and baseline wasm without this proposal.

In any case I wanted to open this up to document my thinking. If anyone else has something they'd like to see before phase 3 please let me know!