Refactoring `CsvFileConnector` and `CsvDataSource`

[sebastian-peter]

What I think this should look like is *Connectors taking care of operations close to the file system, and *DataSource handling higher-level decisions, i.e. where data is stored (paths/databases/etc.). With SqlConnector and SqlDataSource, this division is already in place.

Illustrating the desired change with examples in two areas:

• Different exception handling CsvConnector should then throw (or return within Trys) ConnectorExceptions, and CsvDataSource should throw (or return within Trys) SourceExceptions.
• Separate FileNamingStrategy from CsvFileConnector Currently, CsvFileConnector uses the FileNamingStrategy to determine paths of some entities. This is happening less with #999, and should arguably move to CsvDataSource anyways. CsvFileConnector should not care where a path comes from, it should just deal with it.