style: replace all Unicode with ASCII across entire codebase
All non-ASCII characters replaced with ASCII equivalents:
- Greek letters: mu->u, phi->phi, lambda->lambda, etc.
- Box drawing: single/double -> +, -, |, =
- Math symbols: x (multiply), != (not equal), -> (arrow), etc.
- Superscripts/subscripts: ^2, ^3, _1, _2, etc.
- Emoji/bullets: [OK], [FAIL], [!], *, etc.
- parse_benchmark.py: simplified regex (only ASCII 'us' needed now)
190 files changed. Zero non-ASCII bytes remain in source.
Build verified: library, benchmarks, selftests all pass.
9ffcc985c8