Avoid squirrelly memcpy() call in filesystem.cpp

[StephanTLavavej]

We have an extremely squirrelly line of code that's memcpying two consecutive DWORDs into the beginning of a buffer. There's no reason for this weirdness - it isn't perf-critical, and the optimizer should understand memcpy. Now, code analysis tools (specifically CodeQL) are noticing that this code is a 🐿️ read overrun. Let's avoid this by splitting it up into two separate reads.

For the destination, _Id points to FILE_ID_INFO. Its FileId is FILE_ID_128, which contains BYTE Identifier[16];.

For the source, _Info is BY_HANDLE_FILE_INFORMATION:

typedef struct _BY_HANDLE_FILE_INFORMATION {
  DWORD    dwFileAttributes;
  FILETIME ftCreationTime;
  FILETIME ftLastAccessTime;
  FILETIME ftLastWriteTime;
  DWORD    dwVolumeSerialNumber;
  DWORD    nFileSizeHigh;
  DWORD    nFileSizeLow;
  DWORD    nNumberOfLinks;
  DWORD    nFileIndexHigh;
  DWORD    nFileIndexLow;
} BY_HANDLE_FILE_INFORMATION, *PBY_HANDLE_FILE_INFORMATION, *LPBY_HANDLE_FILE_INFORMATION;

[StephanTLavavej]

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.