added P&R2017 neutral hydrogen mass model

[itrharrison]

Description

Merging will close #234. This implements the Padmanabhan & Refregier HI halo mass model described in that issue.

Checklist

• [X] Follow the Contributor Guidelines
• [X] Write unit tests
• [X] Write documentation strings
• [X] Assign someone from your working team to review this pull request
• [X] Assign someone from the infrastructure team to review this pull request

[itrharrison]

@rrjbca hmm... I'm not sure I agree. HI would be an observable which is fundamentally and conceptually different to the other types of observables we have (galaxies, gravy waves etc) and on a level with them.

We could say that 'halos' implies 'Dark Matter halos'. This model also makes a prediction only for HI mass.

By your argument, could we also, say, not suggest that the gravitational wave merger rates should also have been placed in galaxies?

I can imagine there are things where what you say does make sense though. e.g. multi-wavelength observations of a cluster's gas (HI, X-ray) might require some 'gas' structure within the halos module.

[laurawolz]

I have had similar thoughts, as the HI models I work with a based on galaxies, so. for this case then, HI model could equally fit into the galaxies module.

I think, based on the fact that HI can based on halo and galaxy properties (or possibly other observables), it would make most sense to have it as an own module which can host a variety of HI models.

[rrjbca]

I meant to say that neutral_hydrogen is way too narrow to justify creating an entire new module (in my opinion). If this doesn't belong in halos (probably true) I would at least suggest skypy/gas/hydrogen.py. A new module for gas has a lot more scope to grow and incorporate e.g. ISM, IGM, ICL etc while hydrogen.py can contain all HI properties you care about irrespective of wether it is bound to halos or galaxies and without having to create a new file for each property.

[rrjbca]

Following a discussion we decided that skypy.neutral_hydrogen is a suitable module for this work. This can always be revisited and changed if we find it conflicts with future design choices.