Use the new internal SSSD errors, to simplify error handling.
Instad of having up to 3 different errors to care about (system, krb5 and
pam_status), collapse all error reporting into one error type mapped on
errno_t.

The returned error can contain either SSSD internal errors, kerberos errors
or system errors, they all use differen number spaces so there is no overlap
and tey can be safely merged.

This means the errors bein sent from the child to the parent is not a pam
status error message anymore. The parent properly deals with that.

Also not that this patch removes returning SSS_PAM_SYSTEM_INFO from the
krb5_child for kerberos errors as all it was doing was simply to make the
parent emit the same debug log already emitted by the child, and the code
is simpler if we do not do that.

EIO has always been an odd match, but was used as an error to indicate
that something had gone wrong internally before we had specific SSSD
errors available. Use ERR_INTERNAL instead going forward.

This prevents reportin false errors when internal functions return
a generic EINVAL or EACCES that should just be treated as internal
errors.

This code adds a new range of error codes specific to SSSD,
It also provides helper functions to print out error defintions
like you can do with system error messages and the strerror() function.

The sss_strerror() function can accept both the new sssd errors and
system errno_t errors falling back to the system strerror() if the error
code provide is not a valid SSSD error code.

This bumps the cache major number to 1 as this change is incompatible
with current clients.
The addiotinal key is used to allow name aliases to be added to user
and group entries. Aliases are a string that can be looked up via the
getpwname and getgrnam calls, and are useful to to fetch entries by
alternate names.
Currently only one, optional alias is allowed.

Added new parameter to split_on_separator that allows to skip
empty values.

The whole function was rewritten. Unit test case was added to
check the new implementation.

https://fedorahosted.org/sssd/ticket/1484

We should never try to invalidate an already invalid record as
internal pointers will not be consistent. Carefully test that the
record really is valid when we are fishing for free space, and
properly invalidate records or return a fatal error if something
goes wrong.
In order to make the code more robust always invalidate the whole
data space on initialization by setting all bits to 1, and make sure
to invalidate the whole last allocated slot by converting rec->len to
the number of slots instead of just the space used.

Currently the wildcard lookup '*$' is done before the one for
host/our.hostname@REALM. This means we would ignore a more specific
match in favour of an unspecific match with a principal which is only
used in a AD environment.

I think this is wrong an wildcards should only be used is all specific
lookups fail.

Currently in select_principal_from_keytab() all kind of different
versions of the host principal are looked up in the keytab except for
the plain name the ldap_sasl_authid option. With this patch the plain
name is looked up first.