I was able to get the following code to work using a RAMDirectory, but after
reading the description of the MemoryIndex, I wanted to try to use it instead
for speed reasons. I removed the RAMDir code, and replaced the references with
the MemoryIndex, and all seems to go well till I start steppin
Document.get(FIELD_NAME) will always be null because MemoryIndex does
not store the original untokenized fulltext(s). Those full texts are
thrown away immediately after tokenization (i.e. on addField()),
keeping only the field names and associated (indexed) tokenized
terms. The latter a