Bug#1143886: rust-wide - autopkgtest fails on many architectures.

Package: rust-wide
Version: 1.6.0+ds-1
Severity: serious

The autopkgtests for rust-wide have regressed on many architectures.

On armhf, the compiler is running out of memory, reducing the debuginfo
level may help with this. To reduce the debuginfo level you can use
the following in Cargo.toml

[profile.dev]
debug = 1

On ppc64el, loong64, riscv64 and s390x the tests are failing with

> 294s ---- simd_float::test_fast_round_int stdout ----
> 294s 
> 294s thread 'simd_float::test_fast_round_int' (2384) panicked at tests/simd_float.rs:539:7:
> 294s assertion `left == right` failed
> 294s   left: (0, 0, 1, 1)
> 294s  right: (0, 0, 0, 1)
> 294s stack backtrace:
> 294s    0: __rustc::rust_begin_unwind
> 294s    1: core::panicking::panic_fmt
> 294s    2: core::panicking::assert_failed_inner
> 294s    3: core::panicking::assert_failed
> 294s    4: tests::utils::for_simd_types::for_simd_types_helper
> 294s    5: core::ops::function::FnOnce::call_once
> 294s note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
> 294s 
> 294s T: f32
> 294s N: 4
> 294s 
> 294s ---- simd_float::test_round_int stdout ----
> 294s 
> 294s thread 'simd_float::test_round_int' (2414) panicked at tests/simd_float.rs:587:7:
> 294s 
> 294s expected: (0, 0, 0, 1)
> 294s   actual: (0, 0, 1, 1)
> 294s    value: [0.0, 0.1, 0.5, 0.7]
> 294s stack backtrace:
> 294s    0: __rustc::rust_begin_unwind
> 294s    1: core::panicking::panic_fmt
> 294s    2: tests::utils::for_simd_types::for_simd_types_helper
> 294s    3: core::ops::function::FnOnce::call_once
> 294s note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.

Since it's failing in the same way on a whole bunch of architectures I suspect it's a broken
fallback codepath.