Inconsistent sampling type for Uniform{BigFloat}

[yurivish]

While playing around with the Uniform distribution I noticed a seeming inconsistency between the results of generating a single sample from a high-precision uniform distribution using rand and generating multiple samples.

For a Uniform{BigFloat} distribution, a single sample using rand(dist) yields a BigFloat value while multiple samples using rand(dist, n) yield Float64 values.

julia> using Distributions

julia> dist = Uniform(BigFloat(0), BigFloat(1))
Uniform{BigFloat}(a=0.0, b=1.0)

julia> rand(dist)
0.518440823260704508612661811639554798603057861328125

julia> typeof(rand(dist))
BigFloat

julia> rand(dist, 2)
2-element Array{Float64,1}:
 0.8587350116719372
 0.6892639654172594

julia> typeof(rand(dist, 2))
Array{Float64,1}

julia> versioninfo()
Julia Version 1.5.3
Commit 788b2c77c1 (2020-11-09 13:37 UTC)
Platform Info:
  OS: macOS (x86_64-apple-darwin18.7.0)
  CPU: Intel(R) Core(TM) i9-9980HK CPU @ 2.40GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-9.0.1 (ORCJIT, skylake)