Fix output_list

I have implemented your suggestion and found a bug @pdraper, @matthieu

I had removed the cosmo here since we already test above whether we have cosmo and ages. But if you'd rather keep it, that's fine.

You check if cosmo with redshift or scale factor, but not with the cosmic time (maybe I should change the flag OUTPUT_LIST_AGE to OUTPUT_LIST_COSMIC_TIME?). Here I do not want to compute the scale factor from the cosmic time in a non cosmological simulation. In current master, the code just crash on a segmentation fault in this case.

Continuing the discussion about the header from !560 (merged). I did wonder about defining the type of the file in the parameter file (string instead of a boolean), which would be tidy, but having it in the file is also good documentation, so decided Loic's method was OK. One further thought (which was the part that unsettled me at first glance) was the need for two words with one blank between them, in the header. You could reasonably just look for the first one as sufficient, so then I thought about breaking the header into words and testing those. Getting seriously over engineered... So strcasecmp instead.

The idea of having # is to clearly set it as a comment for most csv reader. In this way, someone can easily read the file with numpy.

assigned to @pdraper

merged

mentioned in commit 11c22cf2

Thanks looks like an improvement, so accepting. When you find a moment I'd like to see a test for this code, if possible, and maybe update the documentation to say that the headers should be written as shown.