cuda::std::simd core functionalities

[fbusato]

Address #30

Description

Implement the core functionalities of C++26 std::simd, reference https://eel.is/c++draft/simd

The PR introduces the minimal set of functionalities that are self-consistent.
The main goal is to provide a C++26-compliant implementation, no specific optimizations have been introduced.

List of features:

• cuda::std::simd namespace and main header.
• ABI tags: simd_abi::fixed_size_simple, simd_abi::native,
• Type traits: alignment, rebind, resize.
• Exposition-only concepts and utilities.
• Flag types: flags, flag_default, flag_convert, flag_aligned, flag_overaligned.
• basic_vec and basic_mask classes.
• fixed_size_simple (custom ABI) specialization.

To do:

• unit test.
• documentation.

[github-actions]

😬 CI Workflow Results

🟥 Finished in 1h 55m: Pass: 82%/99 | Total: 1d 07h | Max: 1h 40m | Hits: 98%/237006

See results here.