OPENDJ-2016 Implement new on disk merge import strategy based on storage engine


The key/value pairs inserted into trees are now rigorously identical to what the indexes are inserting.
Values are merged before final call to Storage.put(). A possible optimization is to start merging values during the first phase.



OnDiskMergeStorageImporter.java:
Added MergingCursor, Consumer, MergingConsumer, NoMultipleValuesConsumer, ImportIDSetsMerger.
In copyTo(), implemented the merge phase.


DefaultIndex.java:
Extracted decodeValue() and toValue() methods.

DN2ID.java:
Extracted toValue().

ImportIDSet.java:
Extracted merge(EntryIDSet) from merge(ImportIDSet).

VLVIndex.java:
Extracted toValue() method.