unyt should not have a single hard-coded metallicity value

[jzuhone]

unyt uses a single value for the conversion from the ratio of metal mass to total mass to the solar value:

https://github.com/yt-project/unyt/blob/eb511a2a2455e1d2d2fbccdbfbff7b142cb5625f/unyt/_physical_ratios.py#L26-L27

Several studies have been carried out of Solar elemental abundances, which have different values. For example, the value used in X-ray astronomy from Anders & Grevesse (1989) is closer to ~0.02. More recent studies have found something similar. Even Cloudy updates its abundance tables as well.

Potential ways to include other options for solar metallicity measurements:

• Allow users to override the definition of Zsun using some function
• Provide a configuration system for unyt and include this as a configurable option (a big task which probably needs to have more reasons for having a configuration system than just solar metallicity).
• Open to other suggestions...

[neutrinoceros]

I agree that this alone doesn't justify adding a configuration system. Supporting a clean way to override a default at runtime seems like a good option.

[brittonsmith]

I would support some sort of set_solar_metallicity function, if that were straightforward to implement. It could perhaps accept a numerical value or one of a list of keywords corresponding to stock values.

[ngoldbaum]

Fixed by #315