downsample?_loop doesn't support tuple

[johnnychen94]

It's a better practice in Julia to represent sizes or parameters as tuple instead of vector/array because 1) Tuple is static and immutable 2) Julia has a lot of optimization on Tuple to reduce its overhead.

For this reason, it's better to change the type annotation of down to Tuple or NTuple{3, Integer}, and give a deprecation for AbstractVector:

MIRT.jl/src/utility/downsample.jl

Lines 100 to 104 in abb2fff

	function downsample2(x::AbstractMatrix{<:Number},
	down::AbstractVector{Int},
	; warn::Bool = isinteractive(),
	T::DataType = eltype(x[1] / down[1])
	)

If you still plan to support AbstractVector, then it could be Union{Tuple, AbstractVector{<:Integer}} in this case. BTW, it's not generic to write AbstractVector{Int}, it should be AbstractVector{<:Integer}

I just accidentally run into this case when setting up benchmark CI, there might be many others in MIRT.

[JeffFessler]

Thanks - I've fixed it and in related places.
There are probably more and I'll look out for them going forward.