dtype of fixed size array with library='pd'

[renyhp]

Hello!

Here is a test ROOT file. test.root.zip
It contains a tree with a custom class holding a C-style array and a std::array.

I am reading it as follows:

f = uproot.open("test.root")
t = f.get("test")
arrays = t.arrays(['arr[2]', 'stlarray[2]'], library='pd')
print(arrays.dtypes) # awkward
print([type(x) for x in arrays.iloc[0, :]]) # awkward.highlevel.Array

Is there a way to make cells hold numpy arrays instead?

EDIT: Maybe I should add: I can do something like

arrays_transformed = arrays.apply(lambda col: col.apply(np.array))
# actually the following also works (to me, it's pretty unintuitive)
# arrays_transformed = arrays.apply(lambda col: col.apply(lambda x: x))
print(arrays_transformed.dtypes) # object
print([type(x) for x in arrays_transformed.iloc[0, :]]) # np.ndarray

which does what I want. However I wonder if there is a more optimized way to do this, ideally with a more specialized call to t.arrays.