[Dynamo] Do not attempt to make nditer spawned arrays writable (#120868)
As they are not, converting `numpy.nditer` to writable is too expensive and tensor values are copied anyway
Minimal reproducer:
```python
import numpy as np
import torch
@torch.compile
def f(x):
return x + 1.0
for x in np.nditer(np.arange(3)):
print(f(x))
```
Fixes https://github.com/pytorch/pytorch/issues/119787
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120868
Approved by: https://github.com/jansel
