Multi-threaded collection parsing

[bflorat]

We are looking for solutions to accelerate jajuk startup. One of them I thought of was the fact of using several threads (2, one for physical items and the second for logical items). Here's the results:

• It basically a bad idea, after tests, I agree with http://lists.apple.com/archives/java-dev/2006/aug/msg00264.html : for batch processing, multi-threading is unefective
• Differences between a single threaded and multi-threaded collection parsing time: single: 10598, 10344, multi: 9917, 9944, 9990
• For collection (file of several Mb only) and only 2 parsers, the fact of parsing different files makes no difference compared with parsing the same file.
• This complexies the collection parsing

As a consequence, we discard this approach.