Dispatch error in `cubic_interp!` for N-D interpolation when N = 1

Hi,

I noticed a dispatch error in cubic_interp! in some N-dimensional code where I set N=1. Here is a mwe:

using FastInterpolations
ndim = 1
nquery = 7
ndata = 5
output = ones(nquery)
grids = ntuple(n -> range(0, 1, length=ndata), ndim)
data = [sum(sin, x) for x in Iterators.product(ntuple(n -> range(0, 1, length=ndata), ndim)...)]
query = [ntuple(n -> y, ndim) for y in range(0, 1, length=nquery)]
cubic_interp!(output, grids, data, query)

While this code runs perfectly fine with ndim=2, the full stacktrace for ndim=1 is

ERROR: MethodError: no method matching cubic_interp!(::Vector{…}, ::StepRangeLen{…}, ::Vector{…}, ::Vector{…})
The function `cubic_interp!` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  cubic_interp!(::AbstractArray, ::AbstractVector{Tg}, ::AbstractVector{Tv}, ::AbstractArray{Tq}; bc, extrap, autocache, deriv, search, hint) where {Tg<:Number, Tv, Tq<:Number}
   @ FastInterpolations ~/.julia/packages/FastInterpolations/me4ge/src/cubic/cubic_oneshot.jl:239
  cubic_interp!(::AbstractArray, ::Tuple{AbstractVector}, ::AbstractVector, ::AbstractArray; coeffs, kwargs...)
   @ FastInterpolations ~/.julia/packages/FastInterpolations/me4ge/src/cubic/nd/cubic_nd_interpolant.jl:178
  cubic_interp!(::AbstractArray, ::NTuple{N, AbstractVector}, ::AbstractArray{<:Any, N}, ::Any; kwargs...) where N
   @ FastInterpolations ~/.julia/packages/FastInterpolations/me4ge/src/cubic/nd/cubic_nd_oneshot.jl:277
  ...

Stacktrace:
 [1] #cubic_interp!#519
   @ ~/.julia/packages/FastInterpolations/me4ge/src/cubic/nd/cubic_nd_interpolant.jl:180 [inlined]
 [2] cubic_interp!(output::Vector{Float64}, grids::Tuple{StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}, data::Vector{Float64}, q::Vector{Tuple{Float64}})
   @ FastInterpolations ~/.julia/packages/FastInterpolations/me4ge/src/cubic/nd/cubic_nd_interpolant.jl:178
 [3] top-level scope
   @ REPL[73]:1
Some type information was truncated. Use `show(err)` to see complete types.

I was using FastInterpolations v0.4.18.

Taking a look at the first method in the stacktrace, it unwraps the grids tuple using only, and the MethodError seems to be caused by not matching the element type of the query points as Numbers due to the fact they were one-element tuples. It seems like applying map(only, queries) would solve the issue, but create unwanted allocations. Perhaps there is a better solution already within the package? Thanks

Start at the N-dimensional dispatch in src/cubic/nd/cubic_nd_interpolant.jl:178 and compare it with the one-dimensional candidate in src/cubic/cubic_oneshot.jl:239. Reproduce the issue with the provided N=1 MWE, then verify that cubic_interp! dispatches successfully without introducing unwanted allocations and that the existing N=2 case still works.