Consider using a shuffling allocator in benchmarks

To separate the performance of code changes from accidental heap object locality, it might make sense to add a feature that enables the [shuffling-allocator](https://docs.rs/shuffling-allocator) crate.

For reference, also see [wasmtime/crates/bench-api/src/lib.rs](https://github.com/bytecodealliance/wasmtime/blob/v47.0.3/crates/bench-api/src/lib.rs#L154-L160).

Start by reading crates/bench-api/src/lib.rs around the referenced benchmark setup and the shuffling-allocator crate documentation. Trace how benchmark features are configured, then add the allocator option so it can be enabled independently; done means benchmarks can run with shuffling enabled without changing the default configuration.

Written by the indexing model from the issue text.