set_bits::fp_traits_non_native warning for long double

[pps83]

#include <boost/multiprecision/cpp_int.hpp>

with intel's compiler for windows I get this:

1>In file included from D:\work-pps\backtest-engine\src\test\Int128Test.cpp:1:
1>In file included from D:\work-pps\boost_1_88_0\boost\multiprecision\cpp_int.hpp:19:
1>In file included from D:\work-pps\boost_1_88_0\boost\multiprecision\number.hpp:12:
1>In file included from D:\work-pps\boost_1_88_0\boost\multiprecision\detail\generic_interconvert.hpp:12:
1>In file included from D:\work-pps\boost_1_88_0\boost\multiprecision\detail\default_ops.hpp:15:
1>In file included from D:\work-pps\boost_1_88_0\boost\multiprecision\detail\fpclassify.hpp:16:
1>In file included from D:\work-pps\boost_1_88_0\boost\math\special_functions\fpclassify.hpp:21:
1>D:\work-pps\boost_1_88_0\boost\math\special_functions\detail\fp_traits.hpp(349,9): : warning : 'memcpy' will always overflow; destination buffer has size 2, but size argument is 4 [-Wfortify-source]
1>  349 |         std::memcpy(reinterpret_cast<unsigned char*>(&x) + 6, &a, 4);

the code that produces the warning:

    static void set_bits(long double& x, uint32_t a)
    {
        std::memcpy(reinterpret_cast<unsigned char*>(&x) + 6, &a, 4);
    }

sizeof(long double) is 8 with the compiler.

https://www.boost.org/doc/libs/1_89_0/boost/math/special_functions/detail/fp_traits.hpp

[pps83]

perhaps that code should have static_assert for long double size