SDFG -> Schedule Tree: Dealiasing fails on array access as a function parameter

Reproducer, with inline comments showing working patterns and the failing one.

Non-urgent as this can be side-stepped.

```python
import numpy as np
import dace

def fn(field_to_copy):
field_to_copy[0, 0] = field_to_copy[0, 1]
field_to_copy[1, 1] = field_to_copy[1, 2]

@dace.program
def my_program(out_field):
# Works
out_field[0, 0, :] = out_field[0, 1, :]
out_field[1, 1, :] = out_field[1, 2, :]
# Works
out_field[0, 0, :] = np.where(out_field[0, 0, :] > 0, out_field[0, 1, :], out_field[0, 1, :])
for k in dace.map[0 : 5]:
# Works
tmp = out_field[:, :, k]
fn(tmp)
# Fails
fn(out_field[:, :, k])

if __name__ == "__main__":
domain = (2, 2, 5)

i_arr = np.ones(domain)
o_arr = np.ones(domain)

sdfg = my_program.to_sdfg(o_arr)

sdfg.as_schedule_tree()

print("Done 🚀")
```