Add pmull-based aarch64 crc32 implementation

[folkertdev]

Blocked on rust-lang/miri#5070, it also needs CI setup.

Based on the assembly in https://github.com/apple-oss-distributions/zlib/blob/06673bf7cb4066003fd14a0b87085c05739fc4ac/AddOn/ZAssembly/crc32lt_arm64.s.

On a mac mini we see

----------------------------------------------------------------------------------------------------
  CHECKSUMS
----------------------------------------------------------------------------------------------------
Benchmark                                  CPython zlib        zlib_py          Speedup
----------------------------------------------------------------------------------------------------
adler32   1 KB                                   0.1 us         0.2 us     1.33x slower
crc32     1 KB                                   0.1 us         0.3 us     2.00x slower
adler32  64 KB                                   3.7 us         2.0 us     1.83x faster
crc32    64 KB                                   2.6 us         2.8 us     1.10x slower
adler32   1 MB                                  57.0 us        30.3 us     1.88x faster
crc32     1 MB                                  28.6 us        41.0 us     1.43x slower
adler32  10 MB                                 568.9 us       301.5 us     1.89x faster
crc32    10 MB                                 286.3 us       405.6 us     1.42x slower

versus

----------------------------------------------------------------------------------------------------
  CHECKSUMS
----------------------------------------------------------------------------------------------------
Benchmark                                  CPython zlib        zlib_py          Speedup
----------------------------------------------------------------------------------------------------
adler32   1 KB                                   0.1 us         0.2 us     1.33x slower
crc32     1 KB                                   0.1 us         0.2 us     1.33x slower
adler32  64 KB                                   3.7 us         2.0 us     1.83x faster
crc32    64 KB                                   2.6 us         1.8 us     1.41x faster
adler32   1 MB                                  57.0 us        30.3 us     1.88x faster
crc32     1 MB                                  28.7 us        26.7 us     1.08x faster
adler32  10 MB                                 568.6 us       301.9 us     1.88x faster
crc32    10 MB                                 286.8 us       265.9 us     1.08x faster

[codecov]

Codecov Report

❌ Patch coverage is 81.30841% with 20 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
zlib-rs/src/crc32/pmull.rs	84.04%	15 Missing ⚠️
zlib-rs/src/crc32.rs	33.33%	4 Missing ⚠️
zlib-rs/src/cpu_features.rs	85.71%	1 Missing ⚠️

Flag	Coverage Δ
fuzz-compress	39.56% <0.00%> (-0.29%)	⬇️
fuzz-decompress	27.26% <0.00%> (-0.06%)	⬇️
test-aarch64-apple-darwin	87.89% <81.73%> (-0.10%)	⬇️
test-aarch64-unknown-linux-gnu	85.33% <81.73%> (-0.05%)	⬇️
test-i686-unknown-linux-gnu	84.98% <0.00%> (-0.03%)	⬇️
test-x86_64-apple-darwin	88.24% <0.00%> (-0.04%)	⬇️
test-x86_64-unknown-linux-gnu	89.41% <0.00%> (-2.99%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
zlib-rs/src/cpu_features.rs	68.18% <85.71%> (+2.39%)	⬆️
zlib-rs/src/crc32.rs	91.08% <33.33%> (-3.71%)	⬇️
zlib-rs/src/crc32/pmull.rs	84.04% <84.04%> (ø)

... and 4 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.