Implement trait HasUnit following SimpleTraits.jl

src/NonlinearFitting.jl

@@ -20,12 +20,13 @@ using ..Collections
 
 export lsqfit
 
-abstract type UnitTrait end
-struct NoUnit <: UnitTrait end
-struct HasUnit <: UnitTrait end
+# This idea is borrowed from [SimpleTraits.jl](https://github.com/mauro3/SimpleTraits.jl/blob/master/src/SimpleTraits.jl).
+abstract type Trait end
+abstract type Not{T<:Trait} <: Trait end
+struct HasUnit <: Trait end
 
-_traitfn(T::Type{<:Number}) = NoUnit
-_traitfn(T::Type{<:AbstractQuantity}) = HasUnit
+_unit_trait(T::Type{<:Real}) = Not{HasUnit}
+_unit_trait(T::Type{<:AbstractQuantity}) = HasUnit
 
 """
     lsqfit(form, eos, xdata, ydata; debug = false, kwargs...)
@@ -48,29 +49,29 @@ function lsqfit(
     kwargs...,
 )
     T = eltype(eos)
-    return lsqfit(form, eos, xdata, ydata, _traitfn(T), kwargs...)
+    return lsqfit(_unit_trait(T), form, eos, xdata, ydata, kwargs...)
 end # function lsqfit
 function lsqfit(
+    ::Type{Not{HasUnit}},
     form::EquationForm,
     eos::EquationOfState,
     xdata::AbstractVector,
-    ydata::AbstractVector,
-    trait::Type{NoUnit};
+    ydata::AbstractVector;
     debug = false,
     kwargs...,
 )
     T = promote_type(eltype(eos), eltype(xdata), eltype(ydata), Float64)
     E = typeof(eos).name.wrapper
     model(x, p) = map(apply(form, E(p...)), x)
-    fitted = curve_fit(model, T.(xdata), T.(ydata), T.(Collections.fieldvalues(eos)); kwargs...)
+    fitted = curve_fit(model, T.(xdata), T.(ydata), T.(Collections.fieldvalues(eos)), kwargs...)
     return debug ? fitted : E(fitted.param...)
 end  # function lsqfit
 function lsqfit(
+    ::Type{HasUnit},
     form::EquationForm,
     eos::EquationOfState,
     xdata::AbstractVector,
-    ydata::AbstractVector,
-    trait::Type{HasUnit};
+    ydata::AbstractVector;
     kwargs...,
 )
     E = typeof(eos).name.wrapper