link(1M) and link(2) should have POSIX behavior

[Garrett D'Amore]

The legacy semantics of link(2) are not needed, except to support link(1M). ln(1M) doesn't have the documented behavior that link(1M) does.

We should eliminate the legacy behavior from the sys call (lib interface really), and the command.

Specifically, if the target (source) of a hard link is a symbolic link, then the link count should be made to the symbolic link's target rather than the link itself. This behaves like AT_SYMLINK_FOLLOW for linkat().

[Garrett D'Amore]

fixes #71 link(1M) and link(2) should have POSIX behavior fixes #72 ln(1) needs -L and -P argument options per XPG7

→ <<cset dd7732652e91>>

[Garrett D'Amore]

fixes #71 link(1M) and link(2) should have POSIX behavior fixes #72 ln(1) needs -L and -P argument options per XPG7

→ <<cset ffa282a46555>>