Unable to compute transpose of arrays with certain data types.

[zoj613]

When using data types other than Complex32, Float32, Float64 and Complex64, The computation of an array's transpose fails with an unsupported operation error.

This isn't an issue with the correctness of the implementation but a weakness of the underlying n-dimensional array library we use: https://github.com/owlbarn/owl.

See the github issue tracking this bug: owlbarn/owl#671

This bug with Owl also affected the storage set_array function until I implemented a workaround that was added in 3523e879ececcde650f9b89e07db11cb7e1bf313

We have two options: To wait until the Owl library fixes this bug, or implement our own array transpose (that will be slower, surely) but very usable.

Note that this issue does not affect codec chains that do not contain that transpose codec.