Padding error when padding width is 0 for antisymmetric and smooth

[miaotianyi]

Hi, this might be rare to come across, but

pywt.pad([1, 2, 3], 0, mode="antisymmetric")
pywt.pad([1, 2, 3], (2, 0), mode="antisymmetric")
pywt.pad([1, 2, 3], ((5, 0),), mode="antisymmetric")
pywt.pad([1, 2, 3], 0, mode="smooth")
pywt.pad([1, 2, 3], (1, 0), mode="smooth")
pywt.pad([1, 2, 3], ((4, 0),), mode="smooth")

Will all cause errors like

Traceback (most recent call last):
  File "", line 63, in <module>
    pywt.pad([1, 2, 3], 0, mode="smooth")
  File "Anaconda3\lib\site-packages\pywt\_dwt.py", line 476, in pad
    xp = np.pad(x, pad_widths, pad_smooth)
  File "<__array_function__ internals>", line 5, in pad
  File "Anaconda3\lib\site-packages\numpy\lib\arraypad.py", line 767, in pad
    function(view[ind], pad_width[axis], axis, kwargs)
  File "Anaconda3\lib\site-packages\pywt\_dwt.py", line 473, in pad_smooth
    vector[-pad_width[1]:] = \
ValueError: could not broadcast input array from shape (0) into shape (3)

Ideally, a padding width of 0 should make pad an identity function. I noticed that only antisymmetric and smooth modes have this issue. Furthermore, the second 0 in the padding width pair seems to be the source of the problem, as pairs like (3, 0) work without errors.