[ITensorsNamedDimsArraysExt] Convert symmetric tensors

[mtfishman]

With this PR, we can now use to_nameddimsarray to convert ITensor, MPS, and MPO objects that contain tensors with the current format for non-fused abelian symmetric ITensors, i.e. an NDTensors.Tensor with BlockSparse storage and QN blocks labels, to the new format we are developing, i.e. a NamedDimsArray wrapping a BlockSparseArray with GradedUnitRange axes. In the current design, the block labels of the graded unit range still use the old format for sector labels, i.e. the ITensors.QuantumNumbers.QN type, but it would be easy to support converting those to the new SymmetrySectors.SectorProduct format. Interestingly, it only required overloading GradedAxes.fuse_labels(::QN, ::QN) to make that work, which I think is a testament to the flexibility of the new code is design!

@ogauthe I also fixed a missing overload for GradedUnitRangeDual that I hit when contracting ITensors with the new data format (similar to the definitions introduced in #1571).

So to be concrete, after this PR you can do the following:

using ITensors: Index, QN, dag, random_itensor
using ITensors.ITensorsNamedDimsArraysExt: to_nameddimsarray
i = Index([QN(0) => 2, QN(1) => 3])
a = random_itensor(i', dag(i))
b = to_nameddimsarray(a)
b' * b ≈ to_nameddimsarray(a' * a)

which returns true. You can see the data format of b is:

julia> using NDTensors: tensor

julia> tensor(b)
5×5 NDTensors.NamedDimsArrays.NamedDimsArray{Float64, 2, NDTensors.BlockSparseArrays.BlockSparseArray{Float64, 2, Matrix{Float64}, NDTensors.SparseArrayDOKs.SparseArrayDOK{Matrix{Float64}, 2, NDTensors.BlockSparseArrays.BlockZero{Tuple{NDTensors.GradedAxes.GradedOneTo{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}, Vector{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}}}, NDTensors.GradedAxes.GradedUnitRangeDual{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}, Vector{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}}, NDTensors.GradedAxes.GradedOneTo{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}, Vector{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}}}}}}}, Tuple{NDTensors.GradedAxes.GradedOneTo{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}, Vector{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}}}, NDTensors.GradedAxes.GradedUnitRangeDual{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}, Vector{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}}, NDTensors.GradedAxes.GradedOneTo{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}, Vector{NDTensors.LabelledNumbers.LabelledInteger{Int64, QN}}}}}}, Tuple{ITensors.IndexID, ITensors.IndexID}} with indices GradedOneTo[QN(0) => 1:2, QN(1) => 3:5]×GradedUnitRangeDual[QN(0) => 1:2, QN(1) => 3:5]:
 -0.47797  -0.059555   0.0       0.0        0.0
 -1.14491  -0.121613   0.0       0.0        0.0
  0.0       0.0        1.17931  -0.128576  -0.615945
  0.0       0.0       -1.16749   0.952272  -0.210146
  0.0       0.0        1.15278  -1.20505    1.12212

To-do:

• Add tests.