Correct sense of crypt(3) NULL checks in init(8) and lock(1)

In r231994, an attempt was made to fix crypt(3) failure returns (NULL).
However, instead of treating crypt(3) failure as authentication failure,
some of the changes treated crypt(3) failure as authentication success.
This is wrong.

r324225 fixed this for ppp, which also inspired this review. The other
changes in the 231994 revision were audited for correctness and look ok.

Reviewed by: jhb
Security: yes
Sponsored by: Dell EMC Isilon
Differential Revision: https://reviews.freebsd.org/D12571