[FEA]: Add support for new extended floating point types

[davebayer]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this request and that I agree to the Code of Conduct

Area

libcu++

Is your feature request related to a problem? Please describe.

CUDA 12.8 introduced several new extended floating point types:

• NVFP4
  • __nv_fp4_e2m2
    • This data format is a 4-bit floating point format with 2 bits for exponent and 1 bit for mantissa. The e2m1 encoding does not support infinity and NaN.
• NVFP6
  • __nv_fp6_e2m3
    • This data format is a 6-bit floating point format with 2 bits for exponent and 3 bits for mantissa. The e2m3 encoding does not support infinity and NaN.
  • __nv_fp6_e3m2
    • This data format is a 6-bit floating point format with 3 bits for exponent and 2 bits for mantissa. The e3m2 encoding does not support infinity and NaN.
• NVFP8
  • __nv_fp8_e8m0
    • This data format is an 8-bit unsigned floating-point format with 8 bits for exponent and 0 bits for mantissa. The ue8m0 encoding does not support infinity. NaN value is limited to 0xff.

Describe the solution you'd like

• Add feature macros for NVFP4 and NVFP6 types, similar to NVFP8
• Specialize cuda::std::numeric_limits for these types
• Implement overloads in <cuda/std/cmath>
• Implement specializations of cuda::std::complex

Describe alternatives you've considered

No response

Additional context

No response