Avoid generated function in generic_exp

[andreasnoack]

The current implementation relies on methods being defined for special number types before

ExponentialUtilities.jl/src/exp.jl

Lines 127 to 135 in fce7ae6

	@generated function exp_pade_p(x, ::Val{k}, ::Val{m}) where {k, m}
	factorial = Base.factorial ∘ big
	p = map(Tuple(0:k)) do j
	num = factorial(k + m - j) * factorial(k)
	den = factorial(k + m) * factorial(k - j)*factorial(j)
	(float ∘ eltype)(x)(num // den) * (x <: Number ? 1 : I)
	end
	:(@evalpoly(x, $(p...)))
	end

is being called, see item 4 in https://docs.julialang.org/en/v1/manual/metaprogramming/#Generated-functions-1. If not then you'll hit a world age error. E.g.

julia> using ExponentialUtilities

julia> f = t -> exp_generic(t)
#1 (generic function with 1 method)

julia> f(0.3)
1.3498588075760032

julia> using ForwardDiff

julia> ForwardDiff.derivative(f, 0.2)
ERROR: MethodError: no method matching ForwardDiff.Dual{ForwardDiff.Tag{var"#1#2",Float64},Float64,1}(::Int64)
The applicable method may be too new: running in world age 27795, while current world is 27806.