Why are cosmological contents often modeled as fluids? Why is it that in cosmology, everything is treated as a "fluid"? What justifies modelling radiation, matter, etc. as "fluids" (even dark energy??)?
When does it break down -- surely at small scales they are no longer "fluid-like"? When people run simulations using these descriptions why is it OK that the fluids don't seem to interact with each other?
 A: All the particles systems can considered approximately as Fluids as long as the mean free time(interaction time) is smaller than typical length scales in the problem.
As an example- You can ask your question to any fluid. When can we take water as fluid. Surely for particles whose size is anywhere close to bond length of $\text{H}_2 \text{O}$ it won't be fluid anymore. 
More relevantly to your question - Imagine sound signal whose wavelength of the order of mean free path of air particles. Then sound vibrations cannot travel effectively and this leads the fluid approximation of air to break down.
In cosmology typical length(or equivalently time) is set by Hubble's constant $H$. Thus to answer your question- Whenever mean free time particles under consideration is smaller than $H^{-1}$ we can use the fluid approximation. 
We use this exact concept to determine Photon decoupling time. When Thompson interaction rate $\Gamma$ becomes equal to $H$ Photons decouple from electrons. 
$$\Gamma = n_e \sigma_T c = H $$
From hereon we cannot take them as a single fluid. However we can take baryons(electrons-protons) still as a single fluid.
Remarks:


*

*Even after that we can do it but we have to be careful about all the detailed interactions between the particles. But when it is small we can blindly take it to be a single fluid with few worries. 

*You have to be careful about the details like comoving coordinates, etc..
