docs: Update example to use "_Float16"

[hpkfft]

Since it is recommended that portable code use the _Float16 type defined by ISO/IEC TS 18661-3:2015, it is probably better to update the example in the nanobind docs accordingly.

[wjakob]

AFAIK _Float16 was added in C++23. It is not part of C++17.

[wjakob]

(Sorry for closing the PR, I got confused between tabs)

[hpkfft]

C++23 added std::float16_t.
The type _Float16 was added to C in the 2015 technical spec (TS) listed above, and C++ recommended that compilers that support both C and C++ support _Float16 for C++ as well.
I think, more importantly, that the existing nanobind example uses __fp16, which is not part of any ISO specification. It is just an example showing how nanobind can support an implementation-dependent data type. In the past, ARM recommended its use. After _Float16 became standardized, ARM recommended that instead. So, I think it would be a kindness to gently steer readers to use _Float16.

[hpkfft]

BTW, the C++23 extended floating point type std::float16_t is intended to be the same as _Float16.

Recommended practice: Any names that the implementation provides for the extended floating-point types described in this subsection that are in addition to the names declared in the <stdfloat> header should be chosen to increase compatibility and interoperability with the interchange types _Float16, _Float32, _Float64, and _Float128 defined in ISO/IEC TS 18661-3 and with future versions of ISO/IEC 9899

[wjakob]

OK, fair enough!