Improve deduplication of parsed names and references
The default matching strategies are to strict for data name and nom. ref. data created by a parser. Also there are some mistakes in the defined matching.
Generally we need to make more often use of EQUAL_OR_SECOND_NULL match mode, because the parsed data is always not very complete while existing data might be more complete. E.g. the authors of a reference might be stored with full name while the full name is usually not available by parsed data. Also place published might be unknown in the parsed data but this might be discussed as it is sometimes part of the parsed string.
We probably need a MatchingStrategyFactory that offers matching strategies specific for parsed data compared to persisted richer data.
There were also wrong matching like not matching null and empty datePublished and taking nomenclaturallyRelevant into account.