Reduce loading size by using .select() on CassandraRDDs

[nstrelow]

By selecting only the needed columns on CassandraRDDs the loading size can be greatly reduced.
e.g. subject_dbpeida all columns: 700MB -> only name: 10MB.

To be used when only certain columns are needed.

Can be applied to blocking and duplicate detection and possibly more.

Probably an extension of the subject class is needed to fit the columns that were selected.