The bounding_box() doesn't apply the order of the axes from the function call to the transcripts

[dariarom94]

Issue occurs in certain case

• points elements contain 3D coordinates {x, y, z}
• query a 2D-cut with {y, x}
• element is transformed

Toy example

```python
from spatialdata.datasets import blobs

df = blobs()["blobs_points"].compute()[["x", "y", "genes"]].copy()
df["z"] = np.random.default_rng(0).uniform(0, 10, len(df))     # add 3D, as real readers emit
pts3d = PointsModel.parse(df, feature_key="genes") # create 
set_transformation(pts3d, Scale([0.5, 2.0], axes=("x", "y")), "global")   # non-Identity
sdata_3d = sd.SpatialData(points={"transcripts": pts3d})

X0, X1, Y0, Y1 = 60, 240, 20, 160
for qaxes, mn, mx in [(["x", "y"], [X0, Y0], [X1, Y1]), (["y", "x"], [Y0, X0], [Y1, X1])]:
    r = sdata_3d.query.bounding_box(axes=qaxes, min_coordinate=mn, max_coordinate=mx,
                                target_coordinate_system="global")
    n = len(r["transcripts"]) if "transcripts" in r.points else "element dropped"
    print(f"axes={str(qaxes):14s} -> {n}")
    r.pl.render_points('transcripts', color="genes",
              size=3).pl.show( colorbar=False, legend_loc=None,
                  title="All transcripts")
```

Output:
axes=['x', 'y'] -> 25
axes=['y', 'x'] -> 16