build: reduce release artifact size

[justinjoy]

Summary

• compile production release targets at Meson optimization level 2 by default
• provide -Dcompact_release=false to honor the caller-selected global optimization
• compare compact O2 artifacts against an explicit O3 control in the main footprint workflow
• document the measurement convention and updated footprint

Size impact

Measured on x86_64 with GCC 16.2.1. Shared libraries are stripped with strip --strip-unneeded; static archives and the CLI are measured as built.

Artifact	O3 control	Compact O2	Reduction
shared library	43,168 B	39,072 B	9.5%
static archive	56,142 B	52,358 B	6.7%
chronoid-gen	37,312 B	33,328 B	10.7%

-Os was smaller but roughly doubled KSUID CLI time, while LTO enlarged the static archive more than threefold. O2 retained throughput in the local paired benchmark.

Validation

• GCC 16.2.1 and Clang 22.1.8 full suites: 23/23 each
• forced scalar on GCC and Clang: 23/23 each
• GCC and Clang no-avx2, no-ssse3, and simd=none: 23/23 each
• clang-tidy: 17 library sources, 0 user-code diagnostics
• release O2, O3 opt-out, debug, debugoptimized, minsize, and explicit O1 flags verified
• exported symbols, SONAME, and runtime dependencies unchanged

Native MSVC, macOS, and ARM coverage is delegated to the existing CI matrix.