Some bonus info about my journalistic work this month:
The homogenous and isotropic assumptions never sat right with me when I first learned about it. Especially after learning about cosmic strings which are theoretical structures that could be as long as the observable universe. couldn't get an explanation that convinced me why we were so sure about it. Interesting to hear that there are experts pushing against it and it's not totally crazy.
To steelman the mainstream here a bit:
Getting anything useful out of inhomogeneous models is hard. The more generally you let spacetime vary, the more parameters you're adding in, and you pretty quickly get to a point where you simply don't have enough evidence to credibly fix those parameters. As a result, most proponents of homogeneity are working with some specific restricted model of what that looks like (for Sarkar and co, it's a dipole). A few (the people I covered earlier) are trying to get more general consequences, but then they don't have as much access to detailed predictions.
And that matters because, in some areas, there are a lot of detailed predictions. The mainstream at this point accepts homogeneity not just because it's convenient, but because they haven't gotten into serious trouble for using it. And that's quite non-trivial at this point, there are a lot of detailed cosmology predictions (variation in the CMB for example) and it's unclear to what extent inhomogeneous models can do the same thing, unless the inhomogeneity is quite small.
(Of course, inhomogeneity advocates will say that the reason it's unclear is because there are so few people working on inhomogeneous cosmology that they haven't had the manpower to figure it out!)
(It's also worth pointing out that for Sarkar et al in particular, the CMB is from their perspective a red herring, because they argue that dark energy doesn't actually have a detectable impact for those observables, or other early-universe ones more generally.)
