Fix reallocarray() on Win32

[kleisauke]

../nip4-9.0.1-2/src/kplot/reallocarray.c:33:16: warning: shift count >= width of type [-Wshift-count-overflow]
   33 |         if ((nmemb >= MUL_NO_OVERFLOW || size >= MUL_NO_OVERFLOW) &&
      |                       ^~~~~~~~~~~~~~~
../nip4-9.0.1-2/src/kplot/reallocarray.c:28:30: note: expanded from macro 'MUL_NO_OVERFLOW'
   28 | #define MUL_NO_OVERFLOW (1UL << (sizeof(size_t) * 4))
      |                              ^  ~~~~~~~~~~~~~~~~~~~~
../nip4-9.0.1-2/src/kplot/reallocarray.c:33:43: warning: shift count >= width of type [-Wshift-count-overflow]
   33 |         if ((nmemb >= MUL_NO_OVERFLOW || size >= MUL_NO_OVERFLOW) &&
      |                                                  ^~~~~~~~~~~~~~~
../nip4-9.0.1-2/src/kplot/reallocarray.c:28:30: note: expanded from macro 'MUL_NO_OVERFLOW'
   28 | #define MUL_NO_OVERFLOW (1UL << (sizeof(size_t) * 4))
      |                              ^  ~~~~~~~~~~~~~~~~~~~~
2 warnings generated.

[jcupitt]

Is that all it was!?!?! I feel silly for not having checked that, I assumed their size maths was OK sigh.

(I hope this hasn't taken too much time off your day job!)

[kleisauke]

That's all. I was wondering why my breakpoint on realloc() wasn’t triggered, and then I noticed this in the logs. 😅