Incorrect replace! of a Tensor in a TensorNetwork with the same Tensor

[jofrevalles]

Summary

The replace! function for a Tensor in a TensorNetwork, where it is replaced with the same Tensor, leads to the deletion of that Tensor:

julia> using Tenet

julia> A = Tensor(rand(2, 2, 2), (:i, :j, :k))
2×2×2 Tensor{Float64, 3, Array{Float64, 3}}: ...

julia> B = Tensor(rand(Float32, 2, 2), (:k, :j))
2×2 Tensor{Float32, 2, Matrix{Float32}}: ...

julia> tn = TensorNetwork([A, B])
TensorNetwork (#tensors=2, #inds=3)

julia> replace!(tn, A => (A))
TensorNetwork (#tensors=1, #inds=2)

julia> tensors(tn) .|> inds
1-element Vector{Vector{Symbol}}:
 [:k, :j]

I think that there might be a conditional in the replace! function that checks whether it's the same Tensor, and this may be causing the problem.

[jofrevalles]

@bsc-quantic/software @starsfordummies @mofeing Also I noticed that quite a lot of the replace! tests for TensorNetworks are commented. This is not a good procedure and I think we should never do that. If a test fails we have to check why, and if we decide that the test is deprecated, we remove it.