CSV writer implementation

[nlaing613]

I know @jjhans built the original but since I pushed the most recent version I thought I'd go ahead and explain the logic of the version I pushed so we can discuss everything.

The main summary is that at a user-selected interval, the primitives, sound-speed, temperature, and cell-centers are written to the file according to a fixed precision. This is meant to replace the previous spatial for-loop that collected the fluid state; I found it slowed down the runs significantly.

The original version computed the heat flux within the csv_writer.py, but I believe this should be linked to the actual boundary layer evaluation; that said, I haven't had the chance to link this. It might also be helpful to add the skin friction coefficient. Regardless, both are omitted in this version. I still like the idea and would like to include them.

@TimothyEDawson, I understand you mentioned changing the file formatting to hdf5 to be compatible with restart files. Is this something we want to pursue further? I probably need to get up to speed with the restart functionality, I admit I haven't used it yet.

[TimothyEDawson]

@nlaing613 @jjhans There are two main things which I'm thinking about, here:

1. I would like to expand and use the variable_info_map idea I introduced in Plotting facilities overhaul #83 where, for each variable you request by name, there is an associated function call which takes in the FluidState and returns the requested variable.
2. The existing FluidState.save/restore functions write and read the FluidState into a much more efficient file format already, and combined with 1. you could use that data to compute whatever you'd like as a post-processing step.

I still think the CSV writer is worthwhile as a runtime output option, similar to the probes. It should be possible to insert more complex functions into the variable_info_map including heat flux and skin friction, and then include them within the CSV output upon request. I'm thinking that the Combustor should generate and retain a variable_info_map within the __init__ method, and then add new methods like heat flux into it when the corresponding source terms are being added to the time integration.

I'll take a pass at adding this functionality soon!