Fix panic in as_array/as_array_mut for arrays with a zero-length dimension

[teddytennant]

NumPy reports zero strides for arrays that have a zero-length axis (e.g. shape [2, 0] → strides (0, 0)). as_view passed these straight into ndarray's from_shape_ptr, whose aliasing check rejects a zero stride on an axis of length > 1 and panics ("The strides must not allow any element to be referenced by two different indices") — even though the array holds no elements and cannot alias anything. This affected both as_array and as_array_mut.

Fix: in as_view's inner, when the shape has size 0, substitute clamped C-contiguous strides (computed from shape[i].max(1)). The data pointer is never dereferenced for an empty array, so the fabricated strides are sound for both immutable and mutable views. Keeps the existing infallible signature (no breaking Result change).

Adds a regression test covering [2, 0] (immutable + mutable), [0, 2], and 3-D [2, 0, 3].

Closes #527.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.