CDS entries with dataclass CON are checked to contain CO lines.
From the documentation for entries with (con)structed sequences is not clear the presence of CO lines is mandatory; see paragraph 3.4.14 in:
ftp://ftp.ebi.ac.uk/pub/databases/embl/doc/usrman.txt
The current validation in EmblEntryReader is setting an error per entry, and for a parsing application it will generate a log for most of the entries resulting in hundreds MB of logs.
This check should be set at a different Severity level or bypassed with ReaderOptions.isIgnoreParserErrors() == true