chore(p256-m): add explicit uint16_t cast in u32_muladd64

[stgloorious]

When splitting 32-bit x and y values into lower and higher 16-bit values, make sure to explicitly cast the higher part to uint16_t as well. This fixes the compiler warning:

src/common/crypto_p256.c: In function ‘u32_muladd64’: src/common/crypto_p256.c:264:29: warning: conversion from ‘uint32_t’ {aka ‘unsigned int’} to ‘uint16_t’ {aka ‘const short unsigned int’} may change value [-Wconversio]
  264 |         const uint16_t xh = x >> 16;
      |                             ^

src/common/crypto_p256.c:265:29: warning: conversion from ‘uint32_t’ {aka ‘unsigned int’} to ‘uint16_t’ {aka ‘const short unsigned int’} may change value [-Wconversio]
  265 |         const uint16_t yh = y >> 16;
      |                             ^