BufferizedNDArray Doesn't Support Matmul

[kylebd99]

Minimal Error Example:

import finchlite
import numpy as np

A = np.ones((2,2))
A = finchlite.asarray(A)
A@A

Unless I've missed something, it seems like this should pass. For example, the alternative version does not error:

import finchlite
import numpy as np

A = np.ones((2,2))
A = finchlite.asarray(A)
finchlite.matmul(A,A)

[mtsokol]

A here is neither LazyTensor nor EagerTensor. We need to add matmul, add, mul, etc. to Tensor base class and dispatch to eager version of finchlite.matmul/add/multiply.

Then A @ A should work regardless whether A is a lazy tensor coming from fl.lazy(arr) or just a tensor as in your case. It's a small tweak then.

[willow-ahrens]

Bufferized should inherit from EagerTensor.