[Torch Dialect] Decompose AtenTriuOp

[Mi-Jiazhi]

decompose like:

import torch

def my_triu(x, diag):
    rows = torch.ops.aten.size(x, -2)
    cols = torch.ops.aten.size(x, -1)

    row_indices = torch.ops.aten.arange(rows).unsqueeze(1)
    col_indices = torch.ops.aten.arange(cols).unsqueeze(0)

    cond = torch.ops.aten.ge(
        col_indices, torch.ops.aten.add(row_indices, diag))
    return torch.ops.aten.where(cond, x, 0)

x = torch.rand(5, 7)
assert torch.allclose(my_triu(x, 0), torch.triu(x, 0))
assert torch.allclose(my_triu(x, 1), torch.triu(x, 1))
assert torch.allclose(my_triu(x, 2), torch.triu(x, 2))
assert torch.allclose(my_triu(x, -1), torch.triu(x, -1))

[qingyunqu]

LGTM

[qingyunqu]

@ramiro050 Could you please review this PR?

[ramiro050]

LGTM, just one request