Add assembly inspection to verify optimization assumptions

Need ability to verify that implementations are using expected optimizations (SIMD, hardware crypto extensions, etc).

Features needed:
- Option to dump generated assembly for critical functions
- Automated detection of instruction usage (SHA extensions, SIMD, etc)
- Report which optimizations are actually being used

This would help validate optimization assumptions. For example, we assume both Zig and Rust use ARM64 SHA extensions, but haven't verified this.

No files, tests, or entry points are named. Start by locating the Zig and Rust benchmark implementations and reviewing how critical functions are built and measured; done means assembly can be inspected, relevant instructions are detected, and the report shows which optimizations are used.

Written by the indexing model from the issue text.