fourier_ratio_deconvolution raises TypeError due to estimate_parameters return value in HyperSpy 2.4.0 and exSpy 0.3.2

[wcdyang]

Describe the bug

Calling EELSSpectrum.fourier_ratio_deconvolution() with the default extrapolate_coreloss=True raises a TypeError: cannot unpack non-iterable bool object. The error occurs in power_law_extrapolation (exspy/signals/eels.py:1291), where pl.estimate_parameters(..., out=True) is expected to return a tuple (A, r) of fitted parameters. However, in HyperSpy 2.4.0, estimate_parameters with out=True returns a bool (success/failure flag) instead. The two packages' APIs seem to be out of sync.

To Reproduce

Steps to reproduce the behavior:

import hyperspy.api as hs
s_ll = hs.load('EELS Spectrum Image (low-loss).dm3')
s_cl = hs.load('EELS Spectrum Image (high-loss).dm3')
s_deconv = s_cl.fourier_ratio_deconvolution(low_loss=s_ll)

Expected behavior

Full traceback:

File .../exspy/signals/eels.py:1036, in EELSSpectrum.fourier_ratio_deconvolution(...)
    cl = self.power_law_extrapolation(window_size=20, extrapolation_size=100)

File .../exspy/signals/eels.py:1291, in EELSSpectrum.power_law_extrapolation(...)
    A, r = pl.estimate_parameters(
        s,
        axis.index2value(axis.size - window_size),
        axis.index2value(axis.size - 1),
        out=True,
    )

TypeError: cannot unpack non-iterable bool object

Python environement:

• eXSpy version: 0.3.2
• HyperSpy version: 2.4.0
• Python version: 3.14.3

Additional context

The workaround is to disable extrapolation:

s_deconv = s_cl.fourier_ratio_deconvolution(
    low_loss=s_ll,
    extrapolate_coreloss=False,
    extrapolate_lowloss=False
)

However, disabling extrapolation causes significant Fourier ringing artefacts in the deconvolved spectrum, making the workaround unsuitable for practical use.

[ericpre]

I have tried with the following example and it works fine:

import exspy
import hyperspy.api as hs

cl, ll = exspy.data.eelsdb(element=["Ti", "Ni"], order="spectrumType")

Ti_cl = cl.remove_background([None, 450.0]).isig[:600.0]
Ti_cl.plot()

Ti_cl_deconv = Ti_cl.fourier_ratio_deconvolution(low_loss=ll)
hs.plot.plot_spectra([Ti_cl, Ti_cl_deconv])

Looking at the code in HyperSpy, when out=True, it will return the parameters:
https://github.com/hyperspy/hyperspy/blob/v2.4.0/hyperspy/_components/power_law.py#L177-L178

If you still have the issue, can you please provide the two files used in your example to check if we can reproduce or provide a reproducible example that doesn't use these files?