I have a different theory about this.  As explained all over the place, on FAT file creation times are good to 10 milliseconds but modification times only good to 2 seconds.  But I can't find one credible word about what the various precisions are for NTFS.  For example,

http://msdn.microsoft.com/en-us/library/windows/desktop/ms724290(v=vs.85).aspx

only gives details about FAT.

Anyway, suppose internal system time is 4.5123 seconds.  On FAT we create a file, and it's recorded as 4.51 seconds.  We immediately modify it, and that's recorded as 4 seconds.  Oops!  The file was modified before it was created :-)

If NTFS similarly records creation times with greater resolution than modification times, then obviously something similar could happen.

Could that explain the test failures?  Possibly.  The file doesn't exist at first, so it's plausible that the initial "modification time" retrieved is really the file creation time.  And that the later modification time really is a modification time.