Improved copying of data structures

I see that there is room to making a nice optimization when copying of data structures. Here is an example:

Test code:

SYSTEMTIME a;
FILETIME b;

void struct_copy()
{
	SYSTEMTIME systemTime;
	FILETIME filetime;
	GetSystemTime(&systemTime);
	SystemTimeToFileTime(&systemTime, &filetime);

	a = systemTime;
	b = filetime;
	
}

HLIL result:

void sub_140001010()

struct SYSTEMTIME systemTime
GetSystemTime(&systemTime)
struct FILETIME fileTime
SystemTimeToFileTime(&systemTime, &fileTime)
struct SYSTEMTIME* rdi = &a
struct SYSTEMTIME* rsi = &systemTime
for (int64_t rcx_2 = 0x10; rcx_2 != 0; rcx_2 = rcx_2 - 1)
  rdi->wYear.b = rsi->wYear.b
  rdi = &rdi->wYear + 1
  rsi = &rsi->wYear + 1
int64_t rax
rax.d = fileTime.dwLowDateTime
rax:4.d = fileTime.dwHighDateTime
b.dwLowDateTime = rax.d
*(&b + 4) = rax:4.d

It would be great if binary ninja could turn these two into a simple = operation.

The issue provides a C example using SYSTEMTIME and FILETIME and shows the resulting HLIL. Start by reproducing this example and inspecting how these structure copies are represented. Done means both copies are emitted as simple assignment operations rather than expanded copy loops or scalar stores.