That is, the given record ids will be treated as records which match
incumbent records, are being compressed into existing data, and so
WILL NOT be imported. The --output-import file will contain records
-whose ids DO NOT occur in --idfile; --output-shelve will contain the
+whose ids DO NOT occur in --idfile; --output-shelved will contain the
records which DO occur.
If '--mode=include' is specified, the reverse occurs.