Saving output fails for some problems

[jpmorgan98]

Reported for Kobyashi type problems including one from the challenge problem. @melekderman reports this happening in the profiler in pure Python mode. I have seen it in Numba (CPU and GPU) with and without MPI on various systems including Dane and Lassen. @alexandermote reported a similar error when there are multiple tallies when doing domain decomposition but any soultion has evaded us. This is happening on v0.11.1 and current dev. Example of an error report

  __  __  ____  __ ____   ____ 
 |  \/  |/ ___|/ /_  _ \ / ___|
 | |\/| | |   /_  / | | | |    
 | |  | | |___ / /| |_| | |___ 
 |_|  |_|\____|// |____/ \____|

           Mode | Numba
      Algorithm | History-based
  MPI Processes | 4
 OpenMP Threads | 1
 Now running TNT...

 Generating output HDF5 files...
Traceback (most recent call last):
  File "/usr/WS1/morgan83/pp-mcdc/kobyashi/highfidelity/input_hf.py", line 78, in <module>
    mcdc.run()
  File "/usr/workspace/morgan83/conda_lassen/envs/prod/lib/python3.12/site-packages/mcdc/main.py", line 91, in run
    generate_hdf5(data, mcdc)
  File "/usr/workspace/morgan83/conda_lassen/envs/prod/lib/python3.12/site-packages/mcdc/main.py", line 1306, in generate_hdf5
    tally_bin = tally_bin.reshape(shape)
                ^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: cannot reshape array of size 60 into shape (3,1,1,1,20,60,100,1,1)

[jpmorgan98]

@melekderman reported that fetching changes from the upstream repo solved the issue

[ilhamv]

This is possibly due to the tally decomposition and recomposition, which are already fixed.

cc: @alexandermote