dask indexing error when stacking series dask arrays

[jmuhlich]

Passing some dask arrays returned by to_dask to dask.array.stack and then calling compute on the result or any slice of it raises an Exception. See reproducer and stack trace below.

I ran into this when testing @tlambert03's bioio_bioformats use-bffile branch. The BioImage.get_dask_stack and get_xarray_dask_stack methods call da.stack internally which leads to the same problem.

import bffile
import dask.array as da

bf = bffile.BioFile('tests/data/s_3_t_1_c_3_z_5.ome.tiff')
bf.open()
istack = da.stack([s.to_dask() for s in bf])
print(istack.shape)
img = istack[0, 0, 0, 0].compute()
print(img.shape)

Traceback (most recent call last):
  File "/home/jmuhlich/development/bffile/.venv/lib/python3.11/site-packages/dask/array/chunk.py", line 430, in getitem
    result = obj[index]
             ~~~^^^^^^^
  File "/home/jmuhlich/development/bffile/src/bffile/_lazy_array.py", line 379, in __getitem__
    key = self._normalize_key(key)
          ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jmuhlich/development/bffile/src/bffile/_lazy_array.py", line 569, in _normalize_key
    raise IndexError(msg)
IndexError: too many indices for array: array is 5-dimensional, but 6 were indexed

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/home/jmuhlich/development/bffile/test1.py", line 8, in <module>
    img = istack[0, 0, 0, 0].compute()
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jmuhlich/development/bffile/.venv/lib/python3.11/site-packages/dask/base.py", line 377, in compute
    (result,) = compute(self, traverse=False, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jmuhlich/development/bffile/.venv/lib/python3.11/site-packages/dask/base.py", line 685, in compute
    results = schedule(expr, keys, **kwargs)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jmuhlich/development/bffile/.venv/lib/python3.11/site-packages/dask/array/chunk.py", line 432, in getitem
    raise ValueError(
ValueError: Array chunk size or shape is unknown. Possible solution with x.compute_chunk_sizes()

[psobolewskiPhD]

Thanks for opening the issue and providing a nice reproducer!
I can reproduce, looks like the placeholder None axis used by dask.stack isn't properly handled.
Probably won't have time to try a fix till the weekend.
I suspect the fix needs to be made in the __get_item__, PR would be welcome!