find_cell fails for a lot of simple input including SPE1 grid generated from flow

The following fails:
```python
from hypothesis import given
import hypothesis.strategies as st

from resdata.grid import Grid
from resdata.grid import GridGenerator

coordinates = st.integers(min_value=1, max_value=30)
increments = st.integers().filter(lambda x: x != 0)

grids = st.builds(
GridGen.create_rectangular,
st.tuples(*([coordinates] * 3)),
st.tuples(*([increments] * 3)),
)

@given(grids)
def test_that_find_cell_finds_the_center_point_of_each_cell(grid):
for cell in grid.cells():
assert cell.valid # succeeds
assert cell.ijk == grid.find_cell(*cell.coordinate) # fails
```

failing example is `GridGenerator.create_rectangular((1, 1, 1), (1, 1, -1))` and test succeeds for
```
coordinates = st.integers(min_value=1, max_value=30)
increments = st.integers(min_value=1, max_value=10000)
```
so this definitely has to do with negative increments.

More worryingly, both the check for valid geometry and find_cell does not work with the SPE1 grid generated from flow:
```python
def test_that_flow_grid_works():
g = Grid(
".../ert/test-data/ert/flow_example/spe1_out/realization-0/iter-0/SPE1.EGRID"
)
for cell in g.cells():
# assert cell.valid # fails
center_point = cell.coordinate
assert g.find_cell(*center_point) == cell.ijk # fails

```