matmul does not support broadcast.

[EikanWang]

The unit test case of test_torch.py failed because matmul does not support broadcast.

Unit test case:

FAIL: test_broadcast_batched_matmul_dpcpp (main.TestTorchDeviceTypeDPCPP)

[EikanWang]

@pinzhenx , could we capture the case if the input tensor needs to be broadcasted, and then prompt a message?

[pinzhenx]

Though we do not support broadcast right now, the issue in this test case is caused by #50

In a call to mm_out, the input sizes were [24, 2] and [2, 1], the size of out was [3, 8], which should be rewritten by mm with new out shape [24, 1]