Reduced chi squared

[Catarina-Alves]

Solves #120, #91, #85 and part of #107 .

I also updated the docstring to reflect what chisq_over_datapoints is.

[tallamjr]

It seems the above build failed due to:

E       AttributeError: module 'snmachine.gps' has no attribute 'extract_GP'

in def test_gps_reduced_chi2() of test_gps_reduced_chi2.py.

Would this be fixed with a change from extract_GP --> compute_gps?

[Catarina-Alves]

It seems the above build failed due to:

E       AttributeError: module 'snmachine.gps' has no attribute 'extract_GP'

in def test_gps_reduced_chi2() of test_gps_reduced_chi2.py.

Would this be fixed with a change from extract_GP --> compute_gps?

Yes, I think it is fixed now.

[Catarina-Alves]

I need to raise an error in get_dict_chisq_over_datapoints_per_label if we don't have labels (eg test dataset).

[tallamjr]

Overall very good. I would say just address the comment from Michelle about raising an error for the case of no labels and that's it really.

In other places where I have asked small questions, no changes are actually required now in this PR.

[tallamjr]

Looks good. I am re-running the Azure pipeline to ensure the CI tests are OK.

Then we will need to run some git commands to update the branch

[tallamjr]

@Catarina-Alves Build 72 shows that the following function is expected by several tests in test/snfeatures_test.py for example:

d = <snmachine.sndata.Dataset object at 0x7f5b08bf9978>, sampler = 'leastsq'
use_redshift = False, nprocesses = 1

    def fit_templates(d, sampler='leastsq', use_redshift=False, nprocesses=1):
        temp_featz = snfeatures.TemplateFeatures(sampler=sampler)
        extr_features = temp_featz.extract_features(d, use_redshift=use_redshift,
                                                    nprocesses=nprocesses, seed=42)
        d.set_model(temp_featz.fit_sn, extr_features)
>       gof = temp_featz.goodness_of_fit(d)
E       AttributeError: 'TemplateFeatures' object has no attribute 'goodness_of_fit'
...

Which I guess was changes in 95a6fff.

Perhaps this just needs to call the new reduced chi^2 functionality instead?