Add tag on DM Annihilation spectral model

[registerrier]

Description

This pull request solves #4056 by simply adding a tag on DarkMatterAnnihilationSpectralModel with short name dm-annihilation.
Note that DM spatial models are not usable in SkyModel as they are not regular SpatialModel.

Dear reviewer

[codecov]

Codecov Report

Merging #4071 (12e96c0) into master (95dc464) will decrease coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #4071      +/-   ##
==========================================
- Coverage   94.04%   94.00%   -0.05%     
==========================================
  Files         162      162              
  Lines       21444    21447       +3     
==========================================
- Hits        20167    20161       -6     
- Misses       1277     1286       +9     

Impacted Files	Coverage Δ
gammapy/astro/darkmatter/spectra.py	98.71% <100.00%> (+0.35%)	⬆️
gammapy/scripts/download.py	84.61% <0.00%> (-15.39%)	⬇️
gammapy/utils/scripts.py	90.47% <0.00%> (-5.18%)	⬇️
gammapy/modeling/models/temporal.py	97.07% <0.00%> (-1.99%)	⬇️
gammapy/maps/wcs/ndmap.py	90.90% <0.00%> (-0.22%)	⬇️
gammapy/catalog/hess.py	97.75% <0.00%> (ø)
gammapy/catalog/gammacat.py	90.75% <0.00%> (ø)
gammapy/maps/region/ndmap.py	93.37% <0.00%> (ø)
gammapy/visualization/utils.py	93.67% <0.00%> (ø)
gammapy/estimators/points/core.py	91.90% <0.00%> (ø)
... and 6 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[adonath]

Thanks @registerrier! My main question is whether adding the tag really solves the problem? The DarkMatterAnnihilationSpectralModel as much more associated information , which needs to be serialized with a custom .to_dict() methods. Including k, z, mass, channel...

[registerrier]

Thanks @registerrier! My main question is whether adding the tag really solves the problem? The DarkMatterAnnihilationSpectralModel as much more associated information , which needs to be serialized with a custom .to_dict() methods. Including k, z, mass, channel...

This is a good point. Technically, some of these quantities could be Parameter objects. But not e.g. channel. How to serialize non Parameter attributes is not clear.
Maybe this is a similar case to NaimaSpectralModel which has no yaml serialization for now.

[adonath]

@registerrier Yes, maybe let's just raise a meaningful error for now?

[registerrier]

I have included additional attributes in the out dictionary. It actually works and allows exporting DMmodels to yaml.

Obviously we will need to look at what should be a Parameter and what is a simple model attribute. But that could be left for later as the DM models will have to be better included in the Model framework (or they will have to be removed if those models are found to be useless).

Opinions @adonath ?

[adonath]

Thanks @registerrier, great to see it works with limited effort. I'll make a more thorough review later...

[adonath]

Thanks @registerrier, I have left one minor comment and one inline suggestion...

[adonath]

Suggested change

	def from_dict(cls, data):
	data = data["spectral"]
	channel = data["channel"]
	mass = u.Quantity(data["mass"])
	jfactor = u.Quantity(data["jfactor"])
	z = data["z"]
	k = data["k"]
	scale = [p["value"] for p in data["parameters"] if p["name"] == "scale"][0]
	return cls(mass, channel, scale=scale, jfactor=jfactor, z=z, k=k)
	def from_dict(cls, data):
	"""Create spectral model from dict
	Parameters
	----------
	data : dict
	DIct with model data
	Returns
	-------
	model : `DarkMatterAnnihilationSpectralModel`
	Dark matter annihilation spectral model
	"""
	data = data["spectral"]
	scale = [p["value"] for p in data.pop("parameters") if p["name"] == "scale"][0]
	return cls(scale=scale, **data)

[adonath]

Maybe add a unit test to establish the quantity here...

[adonath]

What is the status here @registerrier ?

[adonath]

Thanks @registerrier, no further comments from my side.