Spike amplitudes crashes on sorting with no units

Computing the spike amplitudes crashes when giving a sorting with no unit, rather than simply doing nothing.

import spikeinterface.core as si
import spikeinterface.postprocessing as spost

recording = si.load_extractor("recording.bin")
sorting = si.NumpySorting.from_unit_dict({}, 30000)  # Also works with si.load_extractor("sorting").select_units([])

wvf_extractor = si.extract_waveforms(recording, sorting, mode="memory", allow_unfiltered=True)  # Also same error message when using `mode="folder"`
extract waveforms shared_memory multi buffer: 

spost.compute_spike_amplitudes(wvf_extractor)

extract amplitudes:   0%|                                                                                             | 0/1765 [00:00<?, ?it/s]Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/export/home1/users/nsr/wyngaard/dev/spikeinterface/spikeinterface/src/spikeinterface/postprocessing/spike_amplitudes.py", line 171, in compute_spike_amplitudes
    sac.run(**job_kwargs)
  File "/export/home1/users/nsr/wyngaard/dev/spikeinterface/spikeinterface/src/spikeinterface/core/waveform_extractor.py", line 1964, in run
    self._run(**kwargs)
  File "/export/home1/users/nsr/wyngaard/dev/spikeinterface/spikeinterface/src/spikeinterface/postprocessing/spike_amplitudes.py", line 80, in _run
    out = processor.run()
          ^^^^^^^^^^^^^^^
  File "/export/home1/users/nsr/wyngaard/dev/spikeinterface/spikeinterface/src/spikeinterface/core/job_tools.py", line 377, in run
    worker_ctx = self.init_func(*self.init_args)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/export/home1/users/nsr/wyngaard/dev/spikeinterface/spikeinterface/src/spikeinterface/postprocessing/spike_amplitudes.py", line 186, in _init_worker_spike_amplitudes
    worker_ctx["min_shift"] = np.min(peak_shifts)
                              ^^^^^^^^^^^^^^^^^^^
  File "/users/nsr/wyngaard/dev/miniconda3/envs/lussac/lib/python3.11/site-packages/numpy/core/fromnumeric.py", line 2953, in min
    return _wrapreduction(a, np.minimum, 'min', axis, None, out,
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/users/nsr/wyngaard/dev/miniconda3/envs/lussac/lib/python3.11/site-packages/numpy/core/fromnumeric.py", line 88, in _wrapreduction
    return ufunc.reduce(obj, axis, dtype, out, **passkwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: zero-size array to reduction operation minimum which has no identity

Start in spikeinterface/postprocessing/spike_amplitudes.py, especially compute_spike_amplitudes(), _run(), and _init_worker_spike_amplitudes(), where the traceback reaches np.min(peak_shifts). Reproduce with an empty NumpySorting and waveform extractor using the issue’s example. Done means computing spike amplitudes with no units completes without the zero-size reduction error and performs no unnecessary work.