[Docs] Potential typo in cross-referencing Tensor.from_numpy

[MehdiChelh]

Where is the problem?

https://docs.modular.com/max/tutorials/build-custom-ops

What can we do better?

The current documentation states that converting Numpy Arrays to Mojo Tensors can be done using the Max Driver API. However, this creates confusion as the documentation references the Mojo API for this functionality, while the feature appears to be only available through the Python API.

The API's Tensor class is designed for seamless integration with common Python frameworks - it offers zero-copy interoperability with both NumPy arrays and PyTorch tensors. Here's how we can leverage this to create a MAX Tensor from random data...

Specifically, users searching for Tensor.from_numpy in the Mojo API will not find this method, as it exists only in the Python implementation. I encountered this issue myself when attempting to use this conversion feature.

Suggested Improvement: Update the documentation to provide appropriate cross-references to guide users to the correct API documentation.

The API's Tensor class is designed for seamless integration with common Python frameworks - it offers zero-copy interoperability with both NumPy arrays and PyTorch tensors. Here's how we can leverage this to create a MAX Tensor from random data...

Anything else?

No response