Improve error handling [ci patch inc]

[HyperWinX]

Now std::exit and std::unreachable (HyperCPU::unreachable) are replaced by std::abort in debug build, to simplify debugging.

[mentatbot]

I'm getting to work. See my steps and track my cost here.

• Wake on any new activity.
  _{I always see new activity while actively working. When paused, I'll only respond when tagged unless this box is checked.}

[mentatbot]

I've reviewed the PR that improves error handling by replacing std::exit and HyperCPU::unreachable with std::abort in debug builds. This is a good change that will make debugging easier while maintaining the original behavior in release builds.

Key Changes

1. Created a new Exit.hpp header with preprocessor macros:

   • UNREACHABLE() - Uses std::abort() in debug builds, __builtin_unreachable() in release builds
   • EXIT(code) - Uses std::abort() in debug builds, std::exit(code) in release builds
   • ABORT() - Always uses std::abort()

2. Added -D__HCPU_DEBUG flag for debug builds to control this behavior

3. Modified the CPU constructor to take a logger reference instead of creating one internally

   • This is a good design improvement for flexibility and testability

4. Updated test files to accommodate the new behavior:

   • Created HCPU_ASSERT_EXIT macro to use different assertions based on build type

5. Reorganized includes in several files for better clarity

Analysis

The changes look well-implemented and consistent throughout the codebase. Using std::abort() in debug builds is a good practice as it preserves the stack and can trigger a debugger, making it easier to diagnose issues.

In release builds, the behavior remains the same, so there shouldn't be any performance impact or behavior change in production.

The PR appears ready to merge and will improve the development experience when debugging.