Vertical slices of a Grid3D are mislabelled, so isel(x=..).isel(y=..) fails #1047

Selecting a single column out of a dfs3 requires selecting along x and then along y. Both of the obvious ways to write that fail, and the way that works reads wrongly.

da = mikeio.read("tests/testdata/dissolved_oxygen.dfs3")[0]
da.dims                            # ("time", "z", "y", "x")

da.isel(x=56, y=110)               # NotImplementedError: Selecting on multiple
                                   # dimensions in the same call, not yet implemented
da.isel(x=56).isel(y=110)          # IndexError: index 110 is out of bounds
                                   # for axis 0 with size 17
da.isel(x=56).isel(x=110)          # works, and the values are correct

The cause is that a vertical slice is represented as a Grid2D, whose axes are named x and y by definition. After isel(x=56) the original z axis is reported as y and the original y axis as x:

s = da.isel(x=55)
s.dims                             # ("time", "y", "x")
s.geometry.nx, s.geometry.ny       # 112, 17  -- i.e. original ny, nz

So the extents are right and the data is right — isel(x=..).isel(x=..) returns exactly the expected column — but the labels describe a horizontal grid that this object is not. A caller who trusts dims gets an IndexError, and one who works around it has to know that x means two different axes on consecutive lines.

Worth deciding whether a vertical slice should be a Grid2D at all. #1010 covers the related loss of origin/orientation/projection in the same branches; both come from reusing a horizontal grid type for a vertical slice.

Code sites:

• src/mikeio/spatial/_grid_geometry.py (Grid3D.isel, the axis == 1andaxis == 2 branches)
• src/mikeio/dataset/_dataarray.py (DataArray.isel, the multiple-dimension guard)