Three possible bugs in DocumentBatch

[SOLR4189]

Hi,
I started to work with LUWAK for percolator functionality in SOLR. I use ParallelMatcher with SimpleMatcher inside, and DocumentBatch with 3000 docs per batch.

During using LUWAK, I noticed 3 things (maybe bugs):

1. First issue with solution you can read here: Using LeafReader only for first segment #171
   It's relevant for big size batches.

2. Second issue with temp solution and optimal suggested solution you can read here: The lack of hermetics in DocumentBatch #172
   It's relevant for big size batches with different fields between docs

3. Also I noticed when I work with DocumentBatch, I must provide _luwak_id field. It's not a problem, but maybe to metion about it in README file?