missing promotion in gaussjacobi for integer α, β

[stevengj]

This works:

julia> gaussjacobi(500, 0.0, -1/2);

but changing 0.0 to an Int gives an error, probably indicating a missing call to float(α) or similar:

julia> gaussjacobi(500, 0, -1/2)
ERROR: MethodError: no method matching asy1(::Int64, ::Int64, ::Float64, ::Int64)
Closest candidates are:
  asy1(::Integer, ::Float64, ::Float64, ::Integer) at ~/.julia/packages/FastGaussQuadrature/BRLTf/src/gaussjacobi.jl:200
Stacktrace:
 [1] jacobi_asy(n::Int64, α::Int64, β::Float64)
   @ FastGaussQuadrature ~/.julia/packages/FastGaussQuadrature/BRLTf/src/gaussjacobi.jl:183
 [2] gaussjacobi(n::Int64, α::Int64, β::Float64)
   @ FastGaussQuadrature ~/.julia/packages/FastGaussQuadrature/BRLTf/src/gaussjacobi.jl:45
 [3] top-level scope
   @ REPL[98]:1

[dlfivefifty]

jacobi_asy should be restricted to Float64 as the choice of boundary terms is precision dependent:

FastGaussQuadrature.jl/src/gaussjacobi.jl

Line 175 in 8b964b7

function jacobi_asy(n, α, β)

It would be easy to add a promote for the case of differing types of course.

[stevengj]

Maybe promote(float(α), float(β))... since passing integers should be allowed.