confusing phrasing for fp16

[bashbaug]

This description is confusing and should be cleaned up: https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_C.html#cl_khr_fp16

The cl_khr_fp16 extension was promoted to OpenCL C 1.2 as an optional feature, and to OpenCL 3.0 as the optional cl_khr_fp16 feature. The extension provides 16-bit precision scalar and vector floating-point data types and extends many functions to accept these types.

fp16 is still currently an extension and there is no cl_khr_fp16 feature or feature test macro.

The fp64 description below it is a little confusing too and could possibly be tidied up also, depending what we decide to do for fp16.