Update the check for a jagged array during _writeParams

[opotowsky]

What is the change?

Scenario where jagged evaluates to False has a loophole (with py313 and numpy 2.2.1): single-element numpy arrays mixed into a list with non-array entries fails in _writeParams. (Edit: it wasn't a loophole...it's that I was previously using a very old version of numpy that allowed jagged arrays, and so the parameter that ended up as a jagged array slipped through the cracks.)

This updates the check for jagged to account for this scenario by editing _getShape.

Why is the change being made?

I was getting this failure sometimes while running cases (with py313 and numpy 2.2.1):

    JaggedArray(temp, paramDef.name) if jagged else np.array(temp)
                                                    ~~~~~~~~^^^^^^
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (782,) + inhomogeneous part.

This is because the logic for determining whether something is a jagged array misses the scenario where single element numpy arrays are mixed with non array/list entries. The code in _getShape evaluates to (1,) for all these array entries AS WELL AS for int, float, or None, which means the jagged check evaluates to False. Then np.array(temp) fails with the above error. So I have the _getShape helper function return 1 for the latter scenario. Then the set for a list of elements mixed with single element arrays is {(1,), 1}, and causes jagged to be True.

Jagged or not jagged, that is the question

We can nix the above update and keep jagged as False by editing the temp list like so:

temp = [
    t.item() if isinstance(t, np.ndarray) and len(t) == 1 else t
    for t in temp
]

I'm just assuming we want jagged to be true, so went with the solution in the PR. Also, this edits the list after the fact. I don't care either way.

---

Checklist

• This PR has only one purpose or idea.
• Tests have been added/updated to verify any new/changed code.

• The code style follows good practices.
• The commit message(s) follow good practices.

• The release notes have been updated if necessary.
• The documentation is still up-to-date in the doc folder.
• The dependencies are still up-to-date in pyproject.toml.

[opotowsky]

@john-science it doesn't return anything cause all it does is write things to the DB. Towards the bottom of the method you see this:

[john-science]

Okay, one option is to move this private function out to being a regular, private method: Database._getArrayShape:

armi/armi/bookkeeping/db/database.py

Lines 918 to 925 in 85053e3

	def _getShape(arr: [np.ndarray, List, Tuple]):
	"""Get the shape of a np.ndarray, list, or tuple."""
	if isinstance(arr, np.ndarray):
	return arr.shape
	elif isinstance(arr, (list, tuple)):
	return (len(arr),)
	else:
	return (1,)

And then add some unit tests.

class Database:

        @staticmethod
        def _getShape(arr: [np.ndarray, List, Tuple]):

[john-science]

Nice catch!