0.4.0

• A new ML-KEM-768 implementation using verified assembler from the mlkem-native project.

  This drops the dependency on libcrux-ml-kem from rustls-graviola and the crate feature. Runtime performance is +10% on Intel and +42% on ARM64, and rustls-graviola build time drops from 6.6s to 2.6s (including all deps).

• SHA3-256, SHA3-512, SHAKE128 and SHAKE256.

• Improved aarch64 AES-GCM performance thanks to @brian-pane.

What's Changed

• Unit tests for verify.rs RSA-PSS hooks by @brian-pane in #149
• rsa: use new to_pkcs1_der inside to_spki_der by @ctz in #161
• Gate catch_unwind tests with cfg check by @CathalMullan in #150
• Enable debug symbols in benchmark builds by @brian-pane in #162
• Follow new clippy lint for p256 multiplication loop by @ctz in #168
• Fix some additional Clippy warnings. by @brian-pane in #170
• Enable ctgrind on Linux/aarch64 by @brian-pane in #167
• Rework ctgrind/valgrind interaction by @ctz in #174
• Make the zeroing tests more deterministic by @brian-pane in #175
• perf: Add read buffering to wycheproof tests to fix 1-byte reads by @brian-pane in #176
• Upgrade to crabgrind 0.2.x with new API for marking memory uninitialized by @brian-pane in #178
• Use hex crate for CAVP value decoding by @ctz in #179
• Unlock install of cargo-fuzz by @ctz in #180
• Do fewer iterations in rsa_bench when running ctgrind tests. by @brian-pane in #177
• Update s2n-bignum by @ctz in #172
• Implement SHA3-256, SHA3-512, SHAKE128 and SHAKE256 by @ctz in #173
• Add a safety comment to the aarch64 sha3 multiplexer by @brian-pane in #182
• Fix the ctgrind errors in the 25519 tests by @brian-pane in #171
• Do not instrument test modules in coverage by @ctz in #181
• ctgrind fails closed by @ctz in #183
• Fix a ctgrind error on aarch64 by @brian-pane in #184
• Run lints on ARM by @ctz in #185
• Update dependencies by @ctz in #188
• Clarify/extend mlkem768 benchmarks by @ctz in #191
• Stitched AES-GCM for aarch64 by @brian-pane in #165
• Enable the parse-asm scripts to work on macOS by @brian-pane in #190
• ML-KEM-768 implementation by @ctz in #189

Full Changelog: v/0.3.4...v/0.4.0