Make @measure more flexible

[cscherrer]

Made it so things like this now work:

julia> using MeasureTheory

julia> using MacroTools

julia> using StatsFuns

julia> @measure MvNormal(μ,Σ) ≃ (1/sqrt2π^length(μ)) * Lebesgue(Real)^size(μ);

julia> prettify(@macroexpand @measure MvNormal(μ,Σ) ≃ (1/sqrt2π^length(μ)) * Lebesgue(Real)^size(μ))
quote
    struct MvNormal{P} <: MeasureTheory.AbstractMeasure
        par::P
    end
    function MvNormal(nt::NamedTuple)
        P = typeof(nt)
        return MvNormal{P}(nt)
    end
    MvNormal(; kwargs...) = MvNormal((; kwargs...))
    function MeasureTheory.basemeasure(μ::MvNormal{P}) where P
        return (1 / sqrt2π ^ length(μ.par.μ)) * Lebesgue(Real) ^ size(μ.par.μ)
    end
    MvNormal(μ, Σ) = MvNormal(; μ, Σ)
end

julia> μ = randn(3);

julia> Σ = let x = randn(10,3)
           x' * x
       end;

julia> basemeasure(MvNormal(μ, Σ))
MeasureTheory.ScaledMeasure{Float64,PowerMeasure{Lebesgue{Real},1}}(-2.7568155996140185, PowerMeasure{Lebesgue{Real},1}(Lebesgue{Real}(), (3,)))

Currently I'm just escaping the whole thing, we'll need another pass once we have it working properly to make things more hygienic.

[codecov-commenter]

Codecov Report

Merging #18 into master will increase coverage by 6.11%.
The diff coverage is 50.00%.

@@            Coverage Diff             @@
##           master      #18      +/-   ##
==========================================
+ Coverage    8.57%   14.68%   +6.11%     
==========================================
  Files          16       16              
  Lines         140      143       +3     
==========================================
+ Hits           12       21       +9     
+ Misses        128      122       -6     

Impacted Files	Coverage Δ
src/combinators/power.jl	0.00% <0.00%> (ø)
src/combinators/product.jl	0.00% <0.00%> (ø)
src/probability/normal.jl	0.00% <ø> (ø)
src/macros.jl	83.33% <100.00%> (+37.50%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b28d28e...10b2527. Read the comment docs.