`fftn` of zero size arrays

[vtavana]

Following example returns an empty array

#  Name        Version   Build              Channel
# ────────────────────────────────────────────────────
#  numpy       1.26.4   py311h689b997_3  intel  
#  numpy-base  1.26.4   py311h913173e_3  intel
#  mkl_fft  1.3.8    py311h977b55c_72  intel

import numpy
a=numpy.ones((2,0,4))
numpy.fft.fftn(a)
# array([], shape=(2, 0, 4), dtype=complex128)

numpy.fft.fftn(a, axes=(1,))
# array([], shape=(2, 0, 4), dtype=complex128)

while stock NumPy raises a ValueError

#  Name  Version   Build              Channel
# ──────────────────────────────────────────────
#  numpy  1.26.4   py312heda63a1_0  conda-forge

import numpy
a=numpy.ones((2,0,4))
numpy.fft.fftn(a)
# ValueError: Invalid number of FFT data points (0) specified.

numpy.fft.fftn(a, axes=(1,))
# ValueError: Invalid number of FFT data points (0) specified.