combine_layers_ds and tuples with length < 2

[Arjanvw]

When having tuples of length one in "combine_layers", combine_layers_ds throws an error.

Maybe specify in function description or add line of code to handle:
combine_layers = [t for t in combine_layers if len(t) > 1]

[rubencalje]

An example:

extent = [131000, 136800, 471500, 475700]
regis = nlmod.read.get_regis(extent)
combine_layers = [(9, 10)]
regis_combined = nlmod.layers.combine_layers_ds(regis, combine_layers)

performs as expected, while

combine_layers = [(9, )]
regis_combined = nlmod.layers.combine_layers_ds(regis, combine_layers)

gives the following error:

Traceback (most recent call last):

  File ~\Documents\GitHub\nlmod\nlmod\dims\layers.py:670 in combine_layers_ds
    da_dict[k] = da.assign_coords(layer=layer_names)

  File ~\miniconda3\envs\artesia\Lib\site-packages\xarray\core\common.py:661 in assign_coords
    data.coords.update(results)

  File ~\miniconda3\envs\artesia\Lib\site-packages\xarray\core\coordinates.py:567 in update
    self._update_coords(coords, indexes)

  File ~\miniconda3\envs\artesia\Lib\site-packages\xarray\core\coordinates.py:935 in _update_coords
    dims = calculate_dimensions(coords_plus_data)

  File ~\miniconda3\envs\artesia\Lib\site-packages\xarray\core\variable.py:3026 in calculate_dimensions
    raise ValueError(

ValueError: conflicting sizes for dimension 'layer': length 34 on <this-array> and length 33 on {'y': 'y', 'x': 'x', 'layer': 'layer'}

The user does not want to combine any layers, so he can just remove the layer-tuple from the combine_layers-argument. But it would be nice if the error would be more intuitive, or if nlmod would just skip this layer, like you propose.

Also, when combine_layers is empty (which can happen when you remove all tuples from it):

combine_layers = []
regis_combined = nlmod.layers.combine_layers_ds(regis, combine_layers)

Gives the following error

Traceback (most recent call last):

  File ~\Documents\GitHub\nlmod\nlmod\dims\layers.py:636 in combine_layers_ds
    new_top, new_bot, reindexer = layer_combine_top_bot(

  File ~\Documents\GitHub\nlmod\nlmod\dims\layers.py:401 in layer_combine_top_bot
    if i in np.concatenate(combine_layers):

ValueError: need at least one array to concatenate

This should be fixed as well.

[rubencalje]

This was fixed by @dbrakenhoff. Only the case where there were no layers that needed to be combined anymore was not handled. This is now also handled by a70602c

[rubencalje]

Fixed since version 0.11.0