Implement memcpy with misaligned addresses

Implement a C-style `memcpy` that correctly handles misaligned source/destination addresses. ## Task Write a function: ```c void* my_memcpy(void* dst, const void* src, size_t n); ``` It must copy exactly `n` bytes from `src` to `dst` and return `dst`. ## Requirements - Assume `src` and `dst` memory ranges **do not overlap** (i.e., `memcpy` semantics, not `memmove`). - `src` and/or `dst` may be **arbitrarily aligned** (not necessarily word-aligned). - Your implementation should be safe and correct for all `n >= 0`. - Optimize the copy by using wider loads/stores (e.g., `sizeof(size_t)` chunks) when possible, but still handle leading/trailing misaligned bytes correctly. ## Constraints / Notes - You may not call the standard library `memcpy`. - You may assume a typical flat address space and that unaligned access may be slower (or potentially illegal on some platforms), so alignment handling matters.