Add Support for Half-Precision Floating-Point (half Data Type) in DirectX Backend

[code-with-parth]

Description:
Implementing support for the half data type in the DirectX backend can bring significant performance and memory usage benefits for applications that can tolerate reduced precision. Half-precision floating-point operations are widely used in modern graphics applications to optimize shaders and reduce resource consumption.

Proposed Features:

1. Half-Precision Support:

Enable the use of the half data type in shaders translated to the DirectX backend.

Provide functionality for compiling and running shaders using half-precision operations.

2. Performance Gains:

Reduce memory bandwidth by using 16-bit floating-point values instead of 32-bit.

Improve shader execution speed on compatible hardware.

3. Hardware Compatibility:

Ensure compatibility with hardware that supports half precision natively (e.g., modern GPUs).

Include fallback mechanisms for hardware without native support.

Benefits:

Improved performance and reduced memory footprint for shaders.

Alignment with modern graphics API trends that leverage lower-precision data types.

Increased efficiency in memory-constrained environments.

Considerations:

Evaluate precision loss and ensure it doesn’t degrade visual quality.

Provide comprehensive documentation and examples for developers using this feature.

Additional Notes:
I believe this feature would significantly enhance crosstl’s usability and performance. I am happy to discuss this further or contribute to the implementation if required.

[code-with-parth]

@NripeshN Please review this issue and assign it to me if it aligns with the project's goals.