Fix - AArch64 Big Endian Intrinsics

[Jamesbarford]

AArch64 Big Endian Intrinsics

Automatic generation of AArch64 big endian intrinsics with the new generator. The generated intrinsics are conceptually similar to those in arm_neon.h, utilizing simd_shuffle!(...) to reorder bits as required.

Testing was performed using the intrinsic-test program, confirming that the generated big-endian intrinsics produce the same results as their clang counterparts.

This PR includes updates to run.sh and a new Dockerfile. However, I have not integrated these changes into the CI pipeline because building rustc currently requires modifications to memchr - see here.

With the changes in this PR, it should be possible to build memchr again, thereby enabling aarch64_be-unknown-linux-gnu as a Rust target without requiring pre-modifications to memchr during the rustc build process.

Closes: #1484

[rustbot]

r? @Amanieu

rustbot has assigned @Amanieu.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[Amanieu]

LGTM, however there still seems to be some CI failures.

[Jamesbarford]

The pipeline should now be fixed, though it currently fails due to:
unknown import: 'env::llvm.wasm.sub.sat.signed.v16i8' has not been defined.. This appears to be unrelated to this work.

The PR is also significantly smaller, as we've isolated the intrinsics that require simd_shuffle!(...) calls. In the previous commit, many simd_shuffle! calls followed the pattern: simd_shuffle!(a, a, [0, 1, 2, 3, 4, 5, 6, 7, ...]), which, upon inspection, appeared to be no-ops. These have now been removed.

Additionally, I've cleaned up the currently non-functional CI code in run.sh and the aarch64_be-unknown-linux-gnu.

[Jamesbarford]

Looks like the pipeline is fixed by; #1719

I'm not sure how to re-run this pipeline?