[Documentation]: Document Tensor operations that move the Tensor off the symmetric heap

Description of errors

There are PyTorch tensor ops that reallocate the Tensor and hence the Tensor becomes outside of the symmetric heap. We need to document these and possibly offer alternative APIs. For instance:

shmem = iris.iris()
tensor = shmem.zeros(4, 3)  # On symmetric heap
non_contiguous = tensor[::2]  # Non-contiguous slice
reshaped = non_contiguous.reshape(-1)  # Creates copy off heap!

Attach any links, screenshots, or additional evidence you think will be helpful.

No response

Start with the Python APIs shown in the issue: iris.iris(), shmem.zeros(), tensor slicing, and reshape(). Identify tensor operations that reallocate data outside the symmetric heap, then document those cases and any available alternatives. Done means users can recognize these operations and avoid unintentionally leaving the heap.