Conflict with SDL.h when compiled under Cygwin

[Yankes]

When compiling with SDL1 I get error here:

#elif defined(FASTFLOAT_32BIT) || (defined(_WIN64) && !defined(__clang__))
  answer.low = _umul128(a, b, &answer.high); // _umul128 not available on ARM64
#elif defined(FASTFLOAT_64BIT) && defined(__SIZEOF_INT128__)

Exact error:

libs/rapidyaml/c4/ext/fast_float_all.h: In function 'fast_float::value128 fast_float::full_multiplication(uint64_t, uint64_t)':
libs/rapidyaml/c4/ext/fast_float_all.h:420:31: error: invalid conversion from 'uint64_t*' {aka 'long unsigned int*'} to 'long long unsigned int*' [-fpermissive]
  420 |   answer.low = _umul128(a, b, &answer.high); // _umul128 not available on ARM64
      |                               ^~~~~~~~~~~~
      |                               |
      |                               uint64_t* {aka long unsigned int*}
In file included from /usr/include/w32api/intrin.h:41,
                 from libs/rapidyaml/c4/ext/fast_float_all.h:175:
/usr/include/w32api/psdk_inc/intrin-impl.h:1027:85: note:   initializing argument 3 of 'long long unsigned int _umul128(long long unsigned int, long long unsigned int, long long unsigned int*)'
 1027 | unsigned __int64 _umul128(unsigned __int64 a, unsigned __int64 b, unsigned __int64 *hi)
      |                                                                   ~~~~~~~~~~~~~~~~~~^~

overall this can be easy reproduced by:

#include <SDL_opengl.h>
#include "rapidyaml/ryml.hpp"

if I swap order then this magically work.

Overall this could be problem with SDL_opengl, but _umul128 is called by rapidyaml and probably should use types definition form same place where _umul128 is defined. Or at least add exception for cygwin same as was done for clang.
Adding #define __clang__ 1 aside form spam from warnings allow code to compile.