We need to audit the level of concurrency within every benchmark, and ensure the following:
1. That the mode of concurrency is reasonable (e.g. not grossly artificial or needlessly single-threaded)
2. That where applicable the benchmark respects the different concurrency models available via the harness (setting the number of threads etc)
3. That the level and nature of concurrency is documented in the cnf file
Adjusted notes on thread for the cnf files of avrora, batik, eclipse, h2, jython, luindex, pmd, tradebeans, tradesoap and xalan.