:: #jaic_type long_double
The C compiler's long double on the target. Where C makes it the same as double
(Apple arm64, Windows with MSVC) this is float64 itself. Elsewhere it is a 16-byte float in
the C format: x87 extended precision on x86-64 Linux, macOS and MinGW, IEEE binary128 on
arm64 Linux and wasm32. Arithmetic, comparisons and casts run at that full precision;
print shows the value converted to float64.