Also unixify src for symmetry and so we don't have to unixify the
whole path again after join.

fix(log): return singular if only one directory

Options.mode to set mode on copied directories

When comparing two files, one does not save time by comparing the md5
hash values instead of comparing the content directly -- computing the
md5 sum simply takes extra CPU cycles and the entire file is read into
memory anyway.

We now compare the content directly. The entire file is still read
into memory, but we can fix that next.

Before, we were reading both files into memory and comparing the
content. We now use the file-sync-cmp library to only read and compare
small chunks of the files. This means that we can compare large files
and still only use a constant amount of memory.

Improve file comparison

Use grunt function to detect directory