Add aten.unflatten.int support and its torch-to-tosa lowering

[zezhang]

Add aten.unflatten.int op
Add its torch-to-tosa lowering
Update the TorchToTosa/basic.mlir tests

To test e2e tosa lowering:

python -m e2e_testing.main -v -c=tosa

[navahgar]

LGTM

I only see a unit test for TorchToTosa. What about other tests? Any e2e tests expected to pass? Can you confirm that we are missing anything?

[ramiro050]

Can you also add an e2e test similar to the flatten ones:

torch-mlir/python/torch_mlir_e2e_test/test_suite/basic.py

Line 307 in 9b5a4af

class FlattenStaticModule(torch.nn.Module):

[zezhang]

Can you also add an e2e test similar to the flatten ones:

torch-mlir/python/torch_mlir_e2e_test/test_suite/basic.py

Line 307 in 9b5a4af

class FlattenStaticModule(torch.nn.Module):

Just added an e2e test. Note that the aten.unflatten.int op only has TOSA lowering for now, and the e2e test will fail on the Linalg-on-Tensors backend. @ramiro050 Please let me know if you think this is ok.

[ramiro050]

Thanks! Other than a small change request, LGTM