Can we remove `DensityModel`?

[torfjelde]

DensityModel isn't that useful now that we have LogDensityProblems.jl propagated through the ecosystem.

So should we just remove it?

The only downside I see here is that technically LogDensityProblems.jl requires implementing dimension, but this would mean that we can support models with varying dimensionality.

However, as long as the underlying sampler doesn't explicitly call dimension, we could just ignore this tbh 🤷