v0.5.0

Added

• Introduced significant performance optimizations via compiler flags:
  • -Ofast, -march=native, -ffast-math, -fomit-frame-pointer added to release builds (CMake and build script).
  • Enabled Link-Time Optimization (LTO) (-flto) for release builds.
  • Enabled Interprocedural Optimization (IPO) via CMake where supported.
• Added compiler-specific hints for potential performance gains:
  • VSORT_RESTRICT (__restrict__) for pointers.
  • VSORT_LIKELY/VSORT_UNLIKELY (__builtin_expect) for branch prediction.
  • inline and __attribute__((hot)) / VSORT_HOT for critical functions.
• Introduced VSORT_HOT macro to manage __attribute__((hot)) portability.

Changed

• Version bumped to 0.5.0.
• Enabled Apple Silicon optimizations (USE_APPLE_SILICON_OPTIMIZATIONS) by default in CMake.
• Updated and simplified CMake build steps in CI workflow (.github/workflows/apple_silicon_ci.yml).
• Updated build instructions in README.md to reflect current CMake usage.
• Applied optimization flags to the standalone build_script.sh.
• Made internal usage of certain compiler attributes (always_inline, hot) conditional for improved portability.

Fixed

• Improved build portability, especially for compilers like MSVC, by abstracting or conditionally compiling GCC/Clang-specific __attribute__ directives.

Removed

• Outdated // TODO: comments from source code.
• Development roadmap section from README.md.
• Unnecessary forward declarations for swap_int/swap_float in vsort.c.