Consider using a better Uniform sampler

[LilithHafner]

We currently use an affine transformation from [0,1):

Distributions.jl/src/univariate/continuous/uniform.jl

Line 154 in ceb6343

rand(rng::AbstractRNG, d::Uniform) = d.a + (d.b - d.a) * rand(rng)

Perhaps this approach would be better: https://hal.science/hal-03282794 (pdf)

Originally posted by @Tortar in LilithHafner/DynamicDiscreteSamplers.jl#35 (comment)

[quildtide]

It's rather convenient that the author, @goualard-f , even provided a naive Julia implementation that assumes that Distributions,jl is loaded.

It looks to me like all of the overhead of this method can be precomputed in a Sampler, except one of the if-branches.