Error raised by rsp_process() with short input signal

[danibene]

Describe the bug
Given a short respiration signal in which only one peak is detected, a ValueError is raised by rsp_process() when interpolating the amplitude to match the signal length (since there is only one amplitude value corresponding to the single peak detected).

error log

---------------------------------------------------------------------------

ValueError                                Traceback (most recent call last)

[<ipython-input-4-b24b2adbbe1f>](https://localhost:8080/#) in <module>
----> 1 nk.rsp_process(rsp, sampling_rate=sampling_rate)

[/usr/local/lib/python3.7/dist-packages/neurokit2/rsp/rsp_process.py](https://localhost:8080/#) in rsp_process(rsp_signal, sampling_rate, method)
     80     # Get additional parameters
     81     phase = rsp_phase(peak_signal, desired_length=len(rsp_signal))
---> 82     amplitude = rsp_amplitude(rsp_cleaned, peak_signal)
     83     rate = signal_rate(info["RSP_Troughs"], sampling_rate=sampling_rate, desired_length=len(rsp_signal))
     84 

[/usr/local/lib/python3.7/dist-packages/neurokit2/rsp/rsp_amplitude.py](https://localhost:8080/#) in rsp_amplitude(rsp_cleaned, peaks, troughs, interpolation_method)
     75     # Interpolate amplitude to length of rsp_cleaned.
     76     amplitude = signal_interpolate(
---> 77         peaks, amplitude, x_new=np.arange(len(rsp_cleaned)), method=interpolation_method
     78     )
     79 

[/usr/local/lib/python3.7/dist-packages/neurokit2/signal/signal_interpolate.py](https://localhost:8080/#) in signal_interpolate(x_values, y_values, x_new, method, fill_value)
     95     if method == "monotone_cubic":
     96         interpolation_function = scipy.interpolate.PchipInterpolator(
---> 97             x_values, y_values, extrapolate=True
     98         )
     99     else:

[/usr/local/lib/python3.7/dist-packages/scipy/interpolate/_cubic.py](https://localhost:8080/#) in __init__(self, x, y, axis, extrapolate)
    230     """
    231     def __init__(self, x, y, axis=0, extrapolate=None):
--> 232         x, _, y, axis, _ = prepare_input(x, y, axis)
    233         xp = x.reshape((x.shape[0],) + (1,)*(y.ndim-1))
    234         dk = self._find_derivatives(xp, y)

[/usr/local/lib/python3.7/dist-packages/scipy/interpolate/_cubic.py](https://localhost:8080/#) in prepare_input(x, y, axis, dydx)
     44         raise ValueError("`x` must be 1-dimensional.")
     45     if x.shape[0] < 2:
---> 46         raise ValueError("`x` must contain at least 2 elements.")
     47     if x.shape[0] != y.shape[axis]:
     48         raise ValueError("The length of `y` along `axis`={0} doesn't "

ValueError: `x` must contain at least 2 elements.

To Reproduce

import neurokit2 as nk
sampling_rate = 100
duration = 10
rsp = nk.rsp_simulate(sampling_rate=sampling_rate, duration=duration)
nk.rsp_process(rsp, sampling_rate=sampling_rate)

System Specifications

- OS: Linux ( 64bit) 
- Python: 3.7.15 
- NeuroKit2: 0.2.1 

- NumPy: 1.21.6 
- Pandas: 1.3.5 
- SciPy: 1.7.3 
- sklearn: 1.0.2 
- matplotlib: 3.2.2

Expected behaviour
I assume that rather than raising an error, we would like to return the single amplitude value, just repeated so that it matched the signal length.

Suggestion

Modify the following lines in rsp_amplitude() such that if there is only one peak detected, the amplitude value will be repeated:
https://github.com/neuropsychology/NeuroKit/blob/dev/neurokit2/rsp/rsp_amplitude.py#L75-L78

if len(peaks) == 1:
    amplitude = amplitude * np.ones(rsp_cleaned.shape)
else:
    amplitude = signal_interpolate(
        peaks, amplitude, x_new=np.arange(len(rsp_cleaned)), method=interpolation_method
    )

Let me know your thoughts & I'll submit a PR!

[DominiqueMakowski]

I missed this. Yes that's good! I'll add it in #746