In `ForceFieldQhaMaker`, I get `ValueError: There should be only one formula unit` in the `analyze_free_energy` step

[Andrew-S-Rosen]

In running the example in #1411, I get back the following traceback. However, the error message does not provide any useful information to the user, and it's not clear how to proceed.

INFO:jobflow.managers.local:analyze_free_energy failed with exception:
Traceback (most recent call last):
  File "c:\Users\asrosen\AppData\Local\miniconda3\envs\cms\Lib\site-packages\jobflow\managers\local.py", line 117, in _run_job
    response = job.run(store=store)
  File "c:\Users\asrosen\AppData\Local\miniconda3\envs\cms\Lib\site-packages\jobflow\core\job.py", line 629, in run
    response = function(*self.function_args, **self.function_kwargs)
  File "C:\Users\asrosen\Desktop\atomate2\src\atomate2\common\jobs\qha.py", line 160, in analyze_free_energy
    raise ValueError("There should be only one formula unit.")
ValueError: There should be only one formula unit.

[JaGeo]

It means that Phonopy detects different numbers of formula units accross the different volumes. This is inconsistent and would likely lead to different normalozations of the free energy.

I am not sure why this is the case for exactly these settings. Maybe, you can fix the symmetry during optimization?

I am aware this is not easy to debug and not user friendly. Yet, I haven't found a way to solve this robustly and automatically.

[JaGeo]

Or: the supercells are not consistent. However, I thought I implemented a consistent supercell search.

EDIT: yeah, I did...

atomate2/src/atomate2/common/flows/qha.py

Line 146 in 52d3774

if supercell_matrix is None:

I would recommend manual debugging here and looking at the relaxation results