BISICLES solver benchmark executable

Since the vast majority of the CPU time in a typical BISICLES run is spent in the velocity solver, it can be useful to run the solver as a standalone code to examine performance and convergence issues. The code in BISICLES/code/benchmark is designed to read in a BISICLES plotfile written out as a part of a standard BISICLES run and perform the momentum balance velocity solve based on the configuration in the plotfile. In order for this to work, the option

amr.write_solver_rhs = 1

must be set in the input file used by the driver run. Then, the option

main.filename =  hdf5 filename 

must be set in the inputs file used to run the benchmark code, which tells the code where to look for the problem setup. To see how this all works (assuming that Chombo, etc are all set up already), try the following test:

1. cd BISICLES/code/benchmark
2. type "make regression". This causes the following to happen:
   1. If necessary, the driver code in BISICLES/exec2D is compiled
   2. the driver executable is run with BISICLES/benchmark/inputs.regression as the inputs file. In the process, a plotfile named "stream.regression.preSolve.000000.2d.hdf5" is written out. "preSolve" means that it's written immediately before conducting a velocity solve
   3. the solverBenchmark executable is then run using the same inputs file
   4. the code reads in the inputs file and the hdf5 file and performs what should be an identical velocity solve to the one which was performed in the driver.