Fixes #830

[unoebauer]

This PR implements changes to the parsers of the various composition files to address #830.

[unoebauer]

The test_model_reader.py does fail. Judging from the setup of the tests, the modified parser was indeed intended for CMFGEN setups.

Before proceeding, we should discuss our options, @wkerzendorf

[unoebauer]

I would suggest the following:

def read_density_file(filename, filetype):
    file_parsers = {'artis': read_artis_density,
                    'simple_ascii': read_simple_ascii_density,
                    'cmfgen_model': read_cmfgen_density}
    ...
def read_abundances_file(abundance_filename, abundance_filetype,
    file_parsers = {'simple_ascii': read_simple_ascii_abundances,
                    'artis': read_simple_ascii_abundances,
                    'cmfgen_model': read_cmfgen_composition,
                    'simple_csv': read_csv_composition}

def read_csv_isotope_abundance(self, delimiter="\+s", skip_columns=0):
    ...
    for element_symbol_string in df.index[skip_columns:]:
        ...

def read_cmfgen_composition(self, delimiter="\+s"):
    return self.read_csv_isotope_abundance(delimiter=delimiter, skip_columns=4)

def read_csv_composition(self, delimiter="\+s"):
    return self.read_csv_isotope_abundance(delimiter=delimiter, skip_columns=0)

What do you think, @wkerzendorf ?

[wkerzendorf]

I think we should rename the cmfgen thing. I think it should be standardized input that is useful for us and we can make something that converts it to our model. I'd like to chat on that on skype

[unoebauer]

I still think that for now we should go with a quick fix on the lines that I've suggested. The CMFGEN format is quite specific. And standardizing the input is an undertaking that goes beyond the original scope of this PR.

[wkerzendorf]

@unoebauer I think we misunderstand each other - the main thing I'm suggesting is to drop the CMFgen name - as it is not true at the moment. I'll try to catch you online to discuss.

[unoebauer]

The tests should pass now (checked in offline on my laptop). If Travis is happy, I'd say that this PR is ready for merging. What do you think, @wkerzendorf ?

[wkerzendorf]

@unoebauer very nice PR (feature; docu + tests - dass ich das noch erleben darf). Doesn't pass - you need to add the file in the package template

[unoebauer]

I see, I forgot to update the tests in test_tardis_model_density_config.py. I will do that soon.

[unoebauer]

Fixed a few outstanding issues - the complete test suite now passes locally. Hopefully Travis is happy now as well.