Currently the LineReader module is used for reading "records". However, this requires the records to be separated by newlines. Parsing multi-line pica records (see issue #139) requires to split records at a different character. Hence, it would be handy to have a module which can read records from a reader which delimited by an arbitrary character.