Write an example explaining Lucene's segmented architecture

[msfroh]

While I've touched on it in several existing examples, I don't think I've written anything focused on Lucene's segmented architecture.

I think a rough outline would be something like:

1. Add some documents, open reader -- how many segments? 1.
2. Add some more documents, open reader -- now we have 2 segments.
3. How do deletes work? Delete a doc or two from segment 1. Reopen reader (applying deletes). You should still see 2 segments, but now the first segment has a live docs file associated with the current commit generation.
4. Updating documents? Delete document from an old segment and add it in a new segment.
5. Merging - Create a manual MergeSpecification with a couple of OneMerges. Maybe one of them merges two segments while another does a singleton merge to expunge deletes.
6. Merge policies - Explain how merges are usually managed by merge policy, maybe with a brief explanation of TieredMergePolicy.