Decide what to do with text output

[Alomir]

Once #239 is done, what do we want to do with text output from SIPNET? Note that #304 is relevant here as well, particularly in pointing out that our current fixed-width format is always a latent bug waiting to happen.

Some options:

• Keep it as is, perhaps with switching to scientific notation (%g formatting)
• Convert to CSV/TSV
• Convert to "standard" binary

Note that the different options have different impact on downstream tools (eg, PEcAn, CARB?) and internal tests (esp. the smoke tests and tools/smoke_check.py)

[ashiklom]

My vote would be for CSV/TSV. Everything can read it, human readable, no dependencies, easy to implement.

[dlebauer]

If we change the format, I agree with Alexey and would prefer CSV.

I'd be reluctant to change format just to fix a bug. And if range checks or additional space will prevent the error in #304, then I say we keep the current format to avoid downstream disruption (although it appears minor and constrained, to changes to sipnet tool and a few lines in pecan that read the file).

@Alomir I'm not sure what you mean by standard binary - does netcdf count? I also agree with Alexey that writing text should be the default option. Is there another binary format that we should consider?