llvm.aarch64.neon.uaddlp.v8i16.v16i8 is not yet supported

[Eveeifyeve]

Ref: #171

thread 'tokio-rt-worker' (43632566) panicked at /rustc/f5eca4fcfa908d1e038afd19c6e746f075859130/library/core/src/panicking.rs:225:5:
llvm.aarch64.neon.uaddlp.v8i16.v16i8 is not yet supported.
See https://github.com/rust-lang/rustc_codegen_cranelift/issues/171
Please open an issue at https://github.com/rust-lang/rustc_codegen_cranelift/issues
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Backtrace:

* thread #3, name = 'tokio-rt-worker', stop reason = signal SIGABRT
  * frame #0: 0x00000001819d95b0 libsystem_kernel.dylib`__pthread_kill + 8
    frame #1: 0x0000000181a13888 libsystem_pthread.dylib`pthread_kill + 296
    frame #2: 0x0000000181918850 libsystem_c.dylib`abort + 124
    frame #3: 0x00000001013f52bc Discord-Bot`std::sys::pal::unix::abort_internal at mod.rs:305:14 [opt]
    frame #4: 0x00000001013f505c Discord-Bot`std::process::abort at process.rs:2536:5 [opt]
    frame #5: 0x00000001013cef4c Discord-Bot`std::panicking::panic_with_hook at panicking.rs:0 [opt]
    frame #6: 0x00000001013c1030 Discord-Bot`std::panicking::panic_handler::{closure#0} at panicking.rs:691:13 [opt]
    frame #7: 0x00000001013b87c4 Discord-Bot`std::sys::backtrace::__rust_end_short_backtrace::<std::panicking::panic_handler::{closure#0}, !> at backtrace.rs:182:18 [opt]
    frame #8: 0x00000001013c183c Discord-Bot`__rustc::rust_begin_unwind at panicking.rs:689:5 [opt]
    frame #9: 0x00000001013f5978 Discord-Bot`core::panicking::panic_nounwind_fmt::runtime at panicking.rs:122:22 [opt] [inlined]
    frame #10: 0x00000001013f5964 Discord-Bot`core::panicking::panic_nounwind_fmt at mod.rs:2448:9 [opt]
    frame #11: 0x00000001013f5900 Discord-Bot`core::panicking::panic_nounwind at panicking.rs:225:5 [opt]
    frame #12: 0x0000000100d5db04 Discord-Bot`<[u8; 79] as simd_adler32::Adler32Hash>::hash at hash.rs:42:20
    frame #13: 0x0000000100d5ceb4 Discord-Bot`core::core_arch::arm_shared::neon::generated::vmlal_u16 at generated.rs:32097:14
    frame #14: 0x0000000100d5cb0c Discord-Bot`simd_adler32::imp::neon::imp::update at neon.rs:37:16
    frame #15: 0x0000000100d5da9c Discord-Bot`<[u8; 77] as simd_adler32::Adler32Hash>::hash at hash.rs:44:8
    frame #16: 0x0000000100d5c0c0 Discord-Bot`simd_adler32::imp::neon::imp::reduce_add_blocks at neon.rs:115:25
    frame #17: 0x0000000100d55a14 Discord-Bot`<[i16] as core::slice::specialize::SpecFill<i16>>::spec_fill at macros.rs:161:27
    frame #18: 0x0000000100d50dac Discord-Bot`miniz_oxide::inflate::core::decompress + 8900
    frame #19: 0x0000000100d589bc Discord-Bot`<[u8; 8] as core::convert::TryFrom<&[u8]>>::try_from at mod.rs:858:13

[0xmuon]

we can implement this intrinsic in src/intrinsics/llvm_aarch64.rs (likely using per-lane widening add + pairwise reduction, or scalar fallback initially and optimize later)... would love to hear your POV