Analyze the query performance of the Lucene index

ModeShape's query engine is designed to load and serve rows in batches. The current Lucene based index implementation, the first time the query is executed, will preload & cache all the document IDs in memory and serve subsequent batches from batches.

We should analyze how well this approach works for large number of documents and compare against the alternative approach of executing the query on each batch and skipping & collecting only the documents for that batch.