[Users] Empty proto in trusted table causes segfault

[Juha Heinanen]

Script Head writes:

 > Just stumbled on this. Obviously there shouldn't be an empty field but also
 > OpenSER should handle this gracefully.

CVS HEAD code seems to check for null proto values:

		(VAL_TYPE(val) == DB_STRING) && !VAL_NULL(val) &&

-- juha