Possible bug in affine_transform_core

[MFrankenbach]

I noticed that the following call to affine_transform_mpo results in a segmentation fault:

using ITensors
using Quantics

function main()
    A = [ 1 1;
        -1 1 ]
    b = [-32, 32]

    R = 6
    D = 2
    y_sites = [ Index(2, "d=$(d),r=$(r)") for d in 1:D for r in 1:R ]
    x_sites = prime.(y_sites)
        # using "new" indices doesn't help
    # x_sites = [ Index(2, "d=$(d),r=$(r)")' for d in 1:D for r in 1:R ]
    y_sites_mat = reshape(y_sites, R, D)
    x_sites_mat = reshape(x_sites, R, D)
    display(y_sites_mat)
    display(x_sites_mat)
    # CRASHES (segfault)
    trafo_mpo = Quantics.affine_transform_mpo(
        y_sites_mat,
        x_sites_mat,
        A,
        b,
        Quantics.OpenBoundaryConditions()
    )
end

main()

This is due to the use of @inbounds in affine_transform_core. Removing these gives the expected

**ERROR**: BoundsError: attempt to access 5×1×1 Array{Bool, 3} at index [1, 2, 1]

because bitrange has length 1 in this case.

Tentatively changing the line

    bitrange = activebit ? range(0, 1) : range(0, 0)

to

bitrange = range(0,1)

removes the bounds error, but causes the above script to be killed due to excess memory usage (~50 GB) on my Macbook Pro M1.

Julia version: 1.11.7
Branch/commit: main / 7c15521

[MFrankenbach]

After some further experiments, seems like the problem happens when b has negative entries or entries larger than 2^R, e.g. for [-1,1],[-5,5],[100,32]. The power of 2 (32) does not seem to be the problem, as it works with b=[32,32].