Parallelization of temperature and free surface kernels

[GoogleCodeExporter]

What steps will reproduce the problem?
1. increase the number of threads to a value > 1
2. run the simulation
3. compare the results to the one-thread-version

What is the expected output? What do you see instead?

Influence of the number of threads on the simulation result can be seen. 
Probably the barriers are misplaced, and/or boundary conditions or forcing 
terms are still buggy.

Original issue reported on code.google.com by [email protected] on 7 Feb 2011 at 2:02