Add implementation of dpnp.lib.stride_tricks.as_strided

[antonwolfy]

This PR adds a new dpnp.lib namespace and implements dpnp.lib.stride_tricks.as_strided.
The function creates a view into an array with a caller-specified shape and strides, sharing the base array's memory.

It closes #2969.

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpnp/pull/2991/index.html

[coveralls]

coverage: 78.127% (+0.02%) from 78.107% — impl-lib.as_strided into master

[github-actions]

Array API standard conformance tests for dpnp=0.21.0dev2=py313h509198e_19 ran successfully.
Passed: 1373
Failed: 2
Skipped: 5

[vlad-perevezentsev]

Would it be better to raise a warning or an error if the user passes a bool value to check_bounds parameter?

[antonwolfy]

We ignoring that keyword, why do we need to validate it?

[vlad-perevezentsev]

do we need to add lib to submodules as is done for fft/ linalg on line 1060?

[vlad-perevezentsev]

If we pass strides that are not multiples of itemsize we willll get incorrect strides result

In [21]: a = dpnp.array([1, 2, 3, 4, 5], dtype="int32")

In [22]: a_np = dpnp.asnumpy(a)

In [23]: res = dpnp.lib.stride_tricks.as_strided(a, shape=(3,), strides=(6,))

In [24]: res_np = numpy.lib.stride_tricks.as_strided(a_np, shape=(3,), strides=(6,))

In [25]: res.strides
Out[25]: (4,)

In [26]: res_np.strides
Out[26]: (6,)

In [27]: res
Out[27]: array([1, 2, 3], dtype=int32)

In [28]: res_np
Out[28]: array([     1, 196608,      4], dtype=int32)

Maybe it makes sense to add an error exception to ndarray stride logic?