extending MIT krb5 string-to-key API

[Ken Hornstein]

>Since we've cast the existing API into stone now, I won't remove or
>change krb5_c_string_to_key, but it's now the wrong thing for any code
>to call that wants to get the user's real Kerberos key.  Comments?

Seems fine to me.  It occurs to me that even as someone who has seriously
misused to V5 API in the past, I don't think I've ever had a reason to
call krb5_{c_}string_to_key(), so I don't think it will impact _that_
many people.

--Ken