Error when using `in(::T, ::StepRangeLen{T,T,T,Int}) where {T<:CartesianIndex{2}}`

[lkdvos]

When working with LinearAlgebra.diagind(::Diagonal, ::IndexCartesian) I bumped into the following error:

using LinearAlgebra
d = Diagonal(rand(3));
r = diagind(d, IndexCartesian()) # StepRangeLen(CartesianIndex(1, 1), CartesianIndex(1, 1), 3)
CartesianIndex(1,1) in diagind(d, IndexCartesian()) # errors

produces (both on 1.11.2 and nightly)

ERROR: MethodError: no method matching /(::CartesianIndex{2}, ::CartesianIndex{2})
The function `/` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  /(::BigInt, ::BigInt)
   @ Base gmp.jl:517
  /(::BigInt, ::Union{Int16, Int32, Int64, Int8, UInt16, UInt32, UInt64, UInt8})
   @ Base gmp.jl:566
  /(::ComplexF64, ::ComplexF64)
   @ Base complex.jl:397
  ...

Stacktrace:
 [1] _in_range
   @ ./range.jl:1434 [inlined]
 [2] in(x::CartesianIndex{2}, r::StepRangeLen{CartesianIndex{2}, CartesianIndex{2}, CartesianIndex{2}, Int64})
   @ Base ./range.jl:1440

I can fix this in a similar way to base by adding a specialized function:

function _in_range(x::CartesianIndex{N}, r::AbstractRange{CartesianIndex{N}}) where {N}
    isempty(r) && return false
    for i in 1:N
        f, l = first(r)[i], last(r)[i]
        # check for NaN, Inf, and large x that may overflow in the next calculation
        f <= x[i] <= l || l <= x[i] <= f || return false
    end

    i = findfirst(!iszero, step(r).I)
    isnothing(i) && return true

    n = round(Integer, (x[i] - first(r)[i]) / step(r)[i]) + 1
    return n ≥ 1 && n ≤ length(r) && r[n] == x
end

but I wanted to check first if this is the preferred way to do this (and where that function should go)

[mbauman]

Closed by #58616.