Complete the implementation of `dpctl.dptensor.numpy_usm_shared.ndarray.__array_ufunc__`

[diptorupd]

The current implementation of dpctl.dptensor.numpy_usm_shared.ndarray.__array_ufunc__ only implements the __call__ function and not the other needed dunder methods: __reduce__, __reduceat__, __accumulate__, __outer__. These other methods have to be implemented to complete the __array_ufunc__ implementation.

[PokhodenkoSA]

@oleksandr-pavlyk Did you take this issue?

[PokhodenkoSA]

Who could take this issue?

[diptorupd]

@oleksandr-pavlyk do you have bandwidth for this task? If not, can guide someone on what needs to be done?

[diptorupd]

It is still relevant, but unlikely to be fixed as numpy_usm_shared is in maintenance mode and will be deprecated once dpctl.tensor.usm_ndarray is feature complete.

[oleksandr-pavlyk]

numpy_usm_shared has been removed.