Inform users of model fidelity / approximation level

[GiudGiud]

We host models of many kinds, and we try to have the documentation be very clear about what each model does, but we don't really report on:

• what these models can be used for
• how accurate / how approximate each model is
• any large deal-breaking approximation or gotcha (we try to for that one, but I don't think it's systematic)

I suspect this should be easier with the coming documentation tagging capabilities