Define in for CartesianIndex ranges

[jishnub]

Currently, this hits a fallback method that assumes that division is defined for the elements of the range. After this, the following works:

julia> r = StepRangeLen(CartesianIndex(1), CartesianIndex(1), 3);

julia> r[1] in r
true

julia> CartesianIndex(0) in r
false

[Seelengrab]

Ref #57034

This PR seems to reinforce the view that ranges are collections of objects, and not intervals. I'm not sure it's relevant, but I think I remember people being of two minds about this, so I thought I'd mention it.

[jishnub]

For what it's worth, this PR was motivated by failing to compute the intersection of these ranges, which is used in checking for aliasing. I believe intersect treats ranges as collections, as it iterates over them.

Closes #57034, and the suggested implementation in that issue is pretty similar to this one.