Clear unnecessary allocated data when after validator.doValidate().
Validate Stated Relationships file instead of validating Relationships
file twice.
Report errors to a files instead of console.
Report taxonomy errors by effective time.
Allow unpublished components to be imported from FULL RF2 files.

...instead of Strings

Add writeTo(Writer) method to SnomedValidationDefect.

Bulk ID operations will load documents in bulk instead of one by one
before executing the operation.

...instead of referencing variable from anonym class.

This will ensure that only relevant information from the RF2 rows will
be kept in memory and not the entire row.

Component importers will read RF2 rows until they reach the specified
threshold (50000 atm) and then start processing them instead of
processing the rows one by one. This way the importer will execute way
more less index lookups than before (single index query per 50000 rows
to get the current revision).

…ed-rf2-importer-issues

...for the last imported RF2 effective time

Fetch Description objects by storageKey and attach new members to lang
members list directly instead of hacking with low-level revision
objects.