A wrapper around the Apache Commons CSVParser that can recover from malformed data.
Recovery is, at present, rather dumb. CSVParser begins parsing the record anew
from the point where the malformed data appeared.
It would be nice if we could retain the record prefix that has already been parsed
(as well as offsetting data). Unfortunately, these changes all require changes to
CSVParser.getNextRecord(), which relies on private access to Token and Lexer.
(Changed in version 2.8.0) collect has changed. The previous behavior can be reproduced with toSeq.