#7: Fix iteration breakdown, add frequency analysis, and plot dropped nodes

[cwschilly]

Fixes #7
Fixes #8

Also refactors the Python code to be more modular and adds both function timers and a new plot for dropped nodes.

The plot the dropped nodes for previous runs, run from the project dir:

python detection/plot_dropped_nodes.py -s slownode.log -a slownodenanalysis.log -o /path/to/output_dir

Where:

-s <slownode.log>         - The driver output (from slow_node.cc)
-a <slownodeanalysis.log> - The analysis output (from detect_slow_nodes.py)

[cwschilly]

@nlslatt @lifflander Should we ignore the first iteration during analysis?

[nlslatt]

@nlslatt @lifflander Should we ignore the first iteration during analysis?

We should probably do one more than the requested number of iterations and completely throw out the first one. I think the first iteration should not be printed or added to the total time.

[nlslatt]

@cwschilly Are you planning to fix the sensors node name problem as a separate PR or part of this? There are a lot of things that need to be addressed before I can start collecting new data and I'd really like to start collecting it now.

[nlslatt]

It would be helpful if the list of slowest iterations included the iteration number and not just the time.

[cwschilly]

@cwschilly Are you planning to fix the sensors node name problem as a separate PR or part of this? There are a lot of things that need to be addressed before I can start collecting new data and I'd really like to start collecting it now.

@nlslatt I can put everything in this PR. Going to try to get everything fixed up by the end of the day--I'll ping you when it's ready

[nlslatt]

@cwschilly Jonathan had asked me for the barrier after the random initialization (i.e., before the for loop within runBenchmark), which is where I've been using one

[nlslatt]

@cwschilly The node names in the sensors files are still incorrect. All nodes have been given the name of the first node in the allocation.

[nlslatt]

I think the node names are working correctly now.