aten.view's err value too large on integer type

In microsoft/beit-base-patch16-224, there has this pattern

view_13 = torch.ops.aten.view.default(arg223_1, [-1])
index = torch.ops.aten.index.Tensor(arg1_1, [view_13])

where arg1_1.shape = [732, 12] and arg223_1 will to be a list of index

tensor([[731, 729, 729,  ..., 729, 729, 729],
        [730, 364, 363,  ...,   2,   1,   0],
        [730, 365, 364,  ...,   3,   2,   1],
        ...,
        [730, 726, 725,  ..., 364, 363, 362],
        [730, 727, 726,  ..., 365, 364, 363],
        [730, 728, 727,  ..., 366, 365, 364]])

arg223_1 dtype is int64, and value 731 is contained in it, and

if arg223_1 become view_13 by torch.ops.aten.view.default, then value 731 retain 731
if arg223_1 become view_13 by ttnn.reshape, then because ttnn.reshape input dtype must be bfloat16, then value 731 should be converted to bfloat16(731) = 732, and convert back to int64(732) = 732 after ttnn.reshape, so the value 731 become 732 in view_13, then it will casue this error
```
IndexError: index 732 is out of bounds for dimension 0 with size 732
```

One of the suggestion of this issue is NOT lowering torch.ops.aten.view.default to ttnn.reshape if dtype is int64(then beit cannot end-to-end compiled), but I want to get some feedback from TT side

You can reproduce this error by pytest tests/pattern/test_beit_pattern.py

tenstorrent / pytorch2.0_ttnn

aten.view's err value too large on integer type #420